Docstoc

Automated Rapid Prototyping of TUG Specifications Using Prolog for Implementing Atomic Read/ Write Shared Memory in Mobile Ad Hoc Networks

Document Sample
Automated Rapid Prototyping of TUG Specifications Using Prolog for Implementing Atomic Read/ Write Shared Memory in Mobile Ad Hoc Networks Powered By Docstoc
					                                                              (IJCSIS) International Journal of Computer Science and Information Security,
                                                              Vol. 8, No. 2, May 2010




Automated Rapid Prototyping of TUG Specifications
Using Prolog for Implementing Atomic Read/ Write
   Shared Memory in Mobile Ad Hoc Networks.
                                           Fatma Omara# 1, Said El Zoghdy*2, Reham Anwer*3
                                   #
                                       Information Systems and Computers Faculty - Cairo University-Egypt.
                                                     1
                                                         Fatma_omara@hotmail.com

                                            *
                                                Science Faculty – Menufiya University- Egypt.
                                                          2
                                                              Elzoghdy@yahoo.com
                                                      3
                                                          rehamteacher@yahoo.com




Abstract: Rapid prototyping has been used for exploring vague                 conventional software development, operational specification,
user requirements in the front-end of the software life cycle.                rapid prototyping via software transformations, software
Automated rapid prototyping may reduce cost of prototyping                    reuse, and analysis of specifications and programs via testing
and the time of developing it .One automated rapid prototyping                and proofs. The language integrates various software
technique is the direct execution of a specification. Direct                  development paradigms into a coherent whole to fit specific
execution of a specification has the benefits of quick construction
of the prototype, direct support for formal specification, and
                                                                              needs of developing organizations. This language improves
quick response to the specification changes. However existing                 the reusability of formal specifications in the following ways
formal specification languages still have difficulties in specifying          [1]:
software systems such as non functional behavior of the systems.
For non-executable formal specification languages, a prototype                (1) A developer can run a TUG specification as a prototype
may be derived from the specification via software                                to study its behavior due to its executability at the front-
transformations. This approach to rapid prototyping uses a                        end of the software life cycle
formal specification language to automatically generate a
prototype in Prolog via a set of software transformation rules.               (2) A developer can easily write a parametric program
Because there is a direct correspondence between the language                     corresponding to its parametric.
and Prolog, the transformation is mechanical and straight
forward. Specifiers can concentrate on generating the prototype                    The idea of prototyping via Software transformations isn't
without the distraction of transforming one notation into                     new. However, as automatic rapid prototyping approach
another. This formal specification language may not provide
enough abstractions for provide enough abstraction for
                                                                              should avoid a proto- type to be rederived from scratch
prototyping some particular features of systems. Therefore, this              whenever there is a change in the specification. Also, the
approach is designed to support the derived prototype to be                   automated approach should allow developers to easily extend
extended or modified in a modular manner. The specification is                the functions of the prototype manually in case the
written in modules in terms of the language patterns that                     specification language doesn't support abstractions for
support module independence, the prototype is then derived in a               features needed for demonstration. A rapid prototyping
modular way that supports the ease of modifications to the                    approach via Software transformations is presented to achieve
prototype. The software transformation rules used for the                     this goal. User requirement are first written into specification
derivation of prototypes in Prolog are presented. In this paper,              in TUG. The specification needs not necessarily be complete,
we apply this specification on the implementation for atomic
object Read/Write shared memory in mobile ad hoc network.
                                                                              precise and correct corresponding to the user requirements at
                                                                              the first attempt. However, the specification should comply
Keywords: Rapid Prototyping, TUG language, Prolog, Mobile Ad                  with the syntax of the language in order to be further
Hoc Networks.                                                                 processed. Next, a prototype in prolog is automatically
                                                                              derived from the specification. Via software transformations.
                   I. INTRODUCTION
                                                                              The prototype is then exercised by the specifier and the use to
  Tree Unified with Grammar (TUG) was developed to                            clarify the user requirements in the front- end of the Software
support a system to be developed through an integration of                    life cycle.




                                                                          1
                                                                         201                             http://sites.google.com/site/ijcsis/
                                                                                                         ISSN 1947-5500
                                                    (IJCSIS) International Journal of Computer Science and Information Security,
                                                    Vol. 8, No. 2, May 2010



           II. THE GEOQUORUM-APPROACH
  In this paper the GeoQuorums approach has presented for             noted that this approach can be applied to any dynamic
implementing atomic read/write shared memory in mobile ad             network that has a geographic basis. Second, an algorithm is
hoc networks. This approach is based on associating abstract          presented to implement read/write atomic memory using the
atomic objects with certain geographic locations. It is               focal point object model. The implementation of the focal
assumed that the existence of Focal Points, geographic areas          point object model depends on a set of physical regions,
that are normally "populated" by mobile nodes. For example:           known as focal points [2].The mobile nodes within a focal
a focal point may be a road Junction, a scenic observation            point cooperate to simulate a single virtual object, known as a
point [2]. Mobile nodes that happen to populate a focal point         focal point object. Each focal point supports a local broadcast
participate in implementing a shared atomic object, using a           service, LBcast which provides reliable, totally ordered
replicated state machine approach. These objects, which are           broadcast. This service allows each node in the focal point to
called focal point objects, are prone to occasional failures          communicate reliably with every operation completely. The
when the corresponding geographic areas are depopulated.              focal broadcast service is used to implement a type of a
The Geoquorums algorithm uses the fault-prone focal point             replicated state machine, one that tolerates joins and leaves of
objects to implement atomic read/write operations on a fault-         mobile nodes. If a focal point becomes depopulated, then the
tolerant virtual shared object. The Geoquorums algorithm uses         associated focal point object fails. (Note that it doesn't matter
a quorum- based strategy in which each quorum consists of a           how a focal point becomes depopulated, be it as a result of
set of focal point objects. The quorums are used to maintain          mobile nodes failing, leaving the area, going to sleep. etc. Any
the consistency of the shared memory and to tolerate limited          depopulation results in the focal point failing). The
failures of the focal point objects, which may be caused by           Geoquorums algorithm implements an atomic read/write
depopulation of the corresponding geographic areas. The               memory algorithm on top of the geographic abstraction, that
mechanism for changing the set of quorums has presented,              is, on top of the focal point object model. Nodes
thus improving efficiency [2]. Overall, the new Geoquorums            implementing the atomic memory use a Geocast service to
algorithm efficiently implements read/write operations in a           communicate with the focal point objects. In order to achieve
highly dynamic, mobile network. In this chapter, a new                fault tolerance and availability, the algorithm replicates the
approach to designing algorithms for mobile ad hoc networks           read/write shared memory at a number of focal point objects.
is presented. An ad hoc network uses no pre-existing                  In order to maintain consistency, accessing the shared
infrastructure, unlike cellular networks that depend on fixed,        memory requires updating certain sets of focal points known
wired base stations. Instead, the network is formed by the            as quorums. An important aspect of this approach is that the
mobile nodes themselves, which co-operate to route                    members of our quorums are focal point objects, not mobile
communication from sources to destinations. Ad hoc                    nodes. The algorithm uses two sets of quorums (I) get-
communication networks are by nature, highly dynamic.                 quorums (II) put- quorums with property that every get-
Mobile nodes are often small devices with limited energy that         quorum intersects every put-quorum. There is no requirement
spontaneously join and leave the network. As a mobile node            that put-quorums intersect other put-quorums, or get-quorums
moves, the set of neighbors with which at can directly                intersect other get-quorums. The use of quorums allows the
communicate may change completely. The nature of ad hoc               algorithm to tolerate the failure of a limited number of focal
networks makes it challenging to solve the standard problems          point objects. This algorithm uses a Global Position System
encountered in mobile computing, such as location                     (GPS) time service, allowing it to process write operations
management using classical tools. The difficulties arise from         using a single phase, prior single-phase write algorithm made
the lack of a fixed infrastructure to serve as the backbone of        other strong assumptions, for example: relying either on
the network. In this chapter developing a new approach that           synchrony or single writers. This algorithm guarantees that all
allows existing distributed algorithm to be adapted for highly        read operations complete within two phases, but allows for
dynamic ad hoc environments one such fundamental problem              some reads to be completed using a single phase: the atomic
in distributed computing is implementing atomic read/ write           memory algorithm flags the completion of a previous read or
shared memory [2]. Atomic memory is a basic service that              write operation to avoid using additional phases, and
facilitates the implementation of many higher level                   propagates this information to various focal paint objects. As
algorithms. For example: one might construct a location               far as we know, this is an improvement on previous quorum
service by requiring each mobile node to periodically write its       based algorithms. For performance reasons, at different times
current location to the memory. Alternatively, a shared               it may be desirable to use different times it may be desirable
memory could be used to collect real – time statistics. The           to use different sets of get quorums and put-quorums [2]. For
problem of implementing atomic read/write memory is                   example: during intervals when there are many more read
divided into two parts; first, we define a static system model,       operations than write operations, it may be preferable to use
the focal point object model that associates abstract objects         smaller get- quorums that are well distributed, and larger put-
with certain fixed geographic locales. The mobile nodes               quorums that are sparsely distributed. In this case a client can
implement this model using a replicated state machine                 rapidly communicate with a get-quorum while communicating
approach. In this way, the dynamic nature of the ad hoc               with a put – quorum may be slow. If the operational statistics
network is masked by a static model. Moreover it should be            change, it may be useful to reverse the situation.




                                                                  2
                                                               202                              http://sites.google.com/site/ijcsis/
                                                                                                ISSN 1947-5500
                                                               (IJCSIS) International Journal of Computer Science and Information Security,
                                                               Vol. 8, No. 2, May 2010




 A. Mathematical Notation for Geoquorums Approach
      -    I the totally- ordered set of node identifiers.                     and Mobile nodes, two specifications is presented , on for the
      -    i0 є I, a distinguished node identifier in I that is                objects and one for the application running on the mobile
           smaller than all order identifiers in I.                            nodes [2] .
      - S, the set of port identifiers, defined as N>0× OP×I,
                                                                                1) Operation Manager Client: This automaton receives read,
           Where OP= {get, put, confirm, recon- done}.
                                                                               write, and recon requests from clients and manages quorum
      - O, the totally- ordered, finite set of focal point
                                                                               accesses to implement these operations (see fig .3, fig.4, and
           identifiers.
                                                                               fig.5). The Operation Manager (OM) is the collection of all
      - T, the set of tags defined as R ≥0 × I.
                                                                               the operation manager clients (OMi, for all i in I).it is
      - U, the set of operation identifiers, defined as R ≥0 × S.              composed of the focal point objects, each of which is an
      - X, the set of memory locations for each x є X:                         atomic object with the put/get variable type.
         - Vx the set of values for x
         - v0,x є Vx , the initial value of X                                  Signature:
      - M, a totally-ordered set of configuration names                        Input:
      - c0 є M, a distinguished configuration in M that is                     Write (Val)i ,val ∈ V
      smaller than all other names in M.                                       read ( )i
      - C, totally- ordered set of configuration identifies, as                recon (cid)I , cid ∈ C
      defined as:                                                              respond (resp) obj, P, resp ∈ responses ( τ ), obj ∈ O, P = <*,*,
                R ≥0 ×I ×M                                                     i> ∈ S
      - L, set of locations in the plane, defined as R× R                      geo-update (t,L)i , t ∈ R≥0 , L ∈ L
                                                                               output:
            Fig.1Notations Used in the Geoquorums Algorithm.
                                                                               write-ack ( )i
                                                                               read-ack (val)i ,val ∈ V
B. Variable Types for Atomic Read/Write object in                              recon-ack (cid)i , cid ∈ C
Geoquorum Approach for Mobile Ad Hoc Network                                   invoke (inv) obj, P, inv ∈ invocations ( τ ),obj ∈ O, P =
                                                                               <*,*,i> ∈ S
    The specification of a variable type for a read/write object
                                                                               Internal:
in geoquorum approach for mobile ad hoc network is
                                                                               read-2 ( )i
presented and a read/write object has the following variable
                                                                               recon-2 (cid)i , cid ∈ C
type (see figure .2) .
                                                                               State:
Put/get variable type   τ                                                      Confirmed C T, a set of tag ids, initially Ø
                                                                               Conf-id ∈ C, a configuration id, initially <0, i0, c0>
State                                                                          recon- ip, a Boolean flag initially false
Tag   ∈T, initially< 0.i >
                         0
                                                                               Clock ∈ R≥0, a time, initially 0
                                                                               Ongoing - invocations C O× S a set of objects and ports
Value ∈V, initially v    0
                                                                               initially Ø
                                                                               Current-port-number ∈ N >0, used to invoke objects, initially
Config-id ∈C, initially< 0, i , c >
                                 0   0                                         1
Confirmed-set C T, initially Ø                                                 Op, a record with the following components:
                                                                               Type ∈ {read, write, recon}, initially read
Recon-ip, a Boolean, initially false
                                                                               Phase ∈ {{idle, get, put}, initially idle
Operations                                                                     Tag ∈ T, initially <0, i0>
Put (new-tag, new-value, new-config-id)                                        Value ∈ V, initially v0
                                                                               recon-ip , a Boolean flag, initially false
                                                                               Recon-conf-id ∈ C, a configuration id, initially <0, i0, c0>
If (new-tag> tag) then
Value new-value                                                                Acc C 0, a set of data objects, initially Ø
Tag new-tag
                                                                               Fig. 3 Operation Manager Client Signature and State for Node i in I, where
If (new-config-id > config-id) then                                            τ is the Put/Get Variable Type.
Config-id new-config-id
                                                                               Operation Manager Client Transitions
Recon-ip      true
                                                                               Output invoke (<get, config-id>) obj,p
C. Operation Manager                                                           Preconditions:
                                                                               P= <current-port-number, get, i>
   In this section the Operation Manger (OM) is presented, an                  <obj, p> Є ongoing-invocations
algorithm built on the focal/point object Model. As the focal
                                                                               Obj Є op.acc
point Object Model contains two entities, focal point objects




                                                                           3
                                                                          203                                 http://sites.google.com/site/ijcsis/
                                                                                                              ISSN 1947-5500
                                                   (IJCSIS) International Journal of Computer Science and Information Security,
                                                   Vol. 8, No. 2, May 2010

Op.phase =get                                                      Ongoing-invocations
Config-id=conf-id                                                  Ongoing-invocations \ {<obj,p>}
Effect:                                                            Input respond (<confirm-ack>) obj,p
Ongoing-invocations                                                Effect:
Ongoing-invocations U {<obj,p>}                                    Ongoing-invocations
Output invoke (<put, tag, val, config-id>) obj,p                   Ongoing-invocations \ {<obj,p>}
Precondition:                                                      Input respond (<recon-done-ack>) obj,p
P=<current-port-number, put, i>                                    Effect:
<obj,p> Є ongoing-invocations                                      Ongoing-invocations
Obj Є op.acc                                                       Ongoing-invocations \ {<obj,p>}
Op.phase =put
                                                                     Fig.4 Operation Manager Client Invoke/Respond Transitions for node i
tag=op.tag
Val=op.value
Config-id=conf-id                                                  Operation Manager Client Transitions
Effect:                                                            Input write (val)i
Ongoing-invocations              Ongoing-invocations U             Effect:
{<obj, p>}                                                         Current-port-number
Output invoke (<confirm, tag >) obj, p                             Current-port-number +1
Precondition:                                                      Op
P=<k, confirm, i>                                                  <write, put,<clock,i>,Val, recon-ip,<0,i0,c0>, Ø>
<obj,p> Є ongoing-invocations                                      Output write-ack ( ) i
tag ∈ confirmed                                                    Precondition:
Effect:                                                            Conf-id=<time-stamp, pid, c>
Ongoing-invocations                                                If op.recon-ip then
Ongoing-invocations U {<obj,p>}                                    √ C/ ∈ M, э P ∈ put-quorums(C/): P C op.acc
Output invoke (<recon-done, config-id >) obj,p                     Else
Precondition:                                                      Э P ∈ put-quorums(C): P C op.acc
P=<k, recon-done,i>                                                Op.phase=put
<obj,p> Є ongoing-invocations                                      Op.type=write
Recon-ip=false                                                     Effect:
Config-id=conf-id                                                  Op.phase                 idle
Input respond (<get-ack, tag, val, confirmed, new-cid, new-        Confirmed                   confirmed U {op.tag}
rip>) obj,p                                                        Input read ( )i
Effect:                                                            Effect:
If(<current-port-number,get,i>=p)then                              Current-port-number
Op.acc                   op.acc U {obj}                            Current-port-number +1
If (tag>op.tag) then                                               Op                 < read, get,┴,┬, recon-ip, <0,i0,c0>, Ø>
Op.tag                   tag                                       Output read-ack (v)i
Op.value                  val                                      Precondition:
If (new-cid>conf-id) then                                          Conf-id=<time-stamp, pid, c>
Conf-id                  new-cid                                   If op.recon-ip then
Op.recon-ip                    true                                √ C/ ∈ M, э G ∈ get-quorums(C/): G C op.acc
Recon-ip                  new-rip                                  Else
Else if (new-cid=conf-id) then                                     Э G ∈ get-quorums(C): G C op.acc
Recon-ip                  recon-ip Λ new-rip                       Op.phase=get
If(confirm=true)then                                               Op.type=read
Confirmed                 confirmed U {tag}                        Op.tag ∈ confirmed
Ongoing-invocations                                                v= op.value
Ongoing-invocations \ {<obj,p>}                                    Effect:
Input respond (<put-ack, new-cid, new-rip>) obj,p                  Op.phase                  idle
Effect:                                                            Internal read-2( )i
If (<current-port-number, put, i>=p) then                          Precondition:
Op.acc                   op.acc U {obj}                            Conf-id=<time-stamp, pid, c>
If (new-cid>conf-id) then                                          If op.recon-ip then
Conf-id                  new-cid                                   √ C/ ∈ M, э G ∈ get-quorums(C/): G C op.acc
Op.recon-ip                    true                                Else
Recon-ip                  new-rip                                  Э G ∈ get-quorums(C): G C op.acc
Else if (new-cid=conf-id) then                                     Op.phase=get
Recon-ip                  recon-ip Λ new-rip                       Op.type=read




                                                               4
                                                              204                               http://sites.google.com/site/ijcsis/
                                                                                                ISSN 1947-5500
                                                           (IJCSIS) International Journal of Computer Science and Information Security,
                                                           Vol. 8, No. 2, May 2010


        ∈
Op.tag confirmed-set                                                       point (i.e. in the specified physical region) collaborate to
                                                                           implement a focal point object. They take advantage of the
 Effect:
Current-port-number                                                        powerful LBcast service to implement a replicated state
Current-port-number +1                                                     machine that tolerates nodes continually joining and leaving
Op.phase                      put                                          .This replicated state machine consistently maintains the state
Op.recon.ip                recon-ip                                        of the atomic object, ensuring that the invocations are
Op.acc                    Ø                                                performed in a consistent order at every mobile node [3]. In
Output read-ack (v)i                                                       this section an algorithm is presented to implement the focal
Precondition:                                                              point object model. the algorithm allows mobile nodes
Conf-id=<time-stamp, pid, c>                                               moving in and out of focal points, communicating with
If op.recon-ip then                                                        distributed clients through the geocast service, to implement
√ C/ ∈ M, э P ∈ put-quorums(C/): P C op.acc                                an atomic object (with port set q=s)corresponding to a
Else                                                                       particular focal point. We refer to this algorithm as the Focal
Э P ∈ put-quorums(C): P C op.acc                                           Point Emulator (FPE).fig .6 contains the signature and state of
Op.phase=put                                                               the FPE .the code for the FPE client is presented in fig.7. The
Op.type=read                                                               FPE client has three basic purposes. First, it ensures that each
v=op.value                                                                 invocation receives at most one response (eliminating
Effect:                                                                    duplicates).Second, it abstracts away the geocast
Op.phase                 idle                                              communication, providing a simple invoke/respond interface
Confirmed                 confirmed U {op.tag}                             to the mobile node[2] [3]. Third, it provides each mobile node
Input recon (conf-name)i                                                   with multiple ports to the focal point object; the number of
Effect:                                                                    ports depends on the atomic object being implemented. The
Conf-id                  <clock,i, conf-name>                              remaining code for the FPE server is in fig .8.When a node
recon-ip                     true                                          enters the focal point, it broadcasts a join-request message
Current-port-number                                                        using the LBcast service and waits for a response. The other
Current-port-number +1                                                     nodes in the focal point respond to a join-request by sending
Op                < recon, get,┴, ┴, true,conf-id, Ø>                      the current state of the simulated object using the LBcast
Internal recon-2(cid)i                                                     service. As an optimization, to avoid unnecessary message
Precondition                                                               traffic and collisions, if a node observes that someone else has
√ C/ ∈ M, э G ∈ get-quorums(C/): G C op.acc                                already responded to a join-request, and then it does not
√ C/ ∈ M, э P ∈ put-quorums(C/): P C op.acc                                respond. Once a node has received the response to its join-
Op.type=recon                                                              request, then it starts participating in the simulation, by
Op.phase=get                                                               becoming active. When a node receives a Geocast message
cid=op.recon-conf-id                                                       containing an operation invocation, it resends it with the
Effect                                                                     lbcast service to the focal point, thus causing the invocation to
Current-port-number                                                        become ordered with respect to the other LBcast messages
Current-port-number +1                                                     (which are join-request messages, responses to join requests,
Op.phase                   put                                             and operation invocations ).since it is possible that a Geocast
Op.acc                     Ø                                               is received by more than one node in the focal point ,there is
Output recon-ack(c)i                                                       some bookkeeping to make sure that only one copy of the
Precondition                                                               same invocation is actually processed by the nodes. There
cid=op.recon-conf-id                                                       exists an optimization that if a node observes that an
cid= <time-stamp, pid, c>                                                  invocation has already been sent with LBcast service, then it
Э P ∈ put-quorums(C): P C op.acc                                           does not do so. Active nodes keep track of operation
Op.type=recon                                                              invocations in the order in which they receive them over the
Op.phase=put                                                               LBcast service. Duplicates are discarded using the unique
Effect:                                                                    operation ids. The operations are performed on the simulated
If(conf-id=op.recon-conf-id)then                                           state in order. After each one, a Geocast is sent back to the
Recon-ip             false                                                 invoking node with the response. Operations complete when
Op.phase                  idle                                             the invoking node with the response. Operations complete
Input geo-update (t, L) i                                                  when the invoking node remains in the same region as when it
Effect:                                                                    sent the invocation, allowing the geocast to find it. When a
Clock               1                                                      node leaves the focal point, it re-initializes its variables [2]
Fig.5 Operation Manager Client Read/Write/Recon and Geo-update             [3].A subtle point is to decide when a node should start
Transitions for Node                                                       collecting invocations to be applied to its replica of the object
                                                                           state. A node receives a snapshot of the state when it joins.
D. Focal Point Emulator Overview
                                                                           However by the time the snapshot is received, it might be out
  The focal point emulator implements the focal point object               of date, since there may have been some intervening messages
Model in an ad hoc mobile network. The nodes in a focal                    from the LBcast service that have been received since the




                                                                       5
                                                                      205                             http://sites.google.com/site/ijcsis/
                                                                                                      ISSN 1947-5500
                                                                (IJCSIS) International Journal of Computer Science and Information Security,
                                                                Vol. 8, No. 2, May 2010

snapshot was sent. Therefore the joining node must record all                    Answered-join-reqs set of ids of Join requests that have
the operation invocations that are broadcast after its join                      already been answered, initially Ø.
request was broadcast but before it received the snapshot .this                  val ∈ V, holds current value of the simulated atomic object,
is accomplished by having the joining node enter a "listening"                   initially v0.
state once it receives its own join request message; all
invocations received when a node is in either the listening or                   Pending-ops, queue of operations waiting to be simulated,
the active state are recorded, and actual processing of the                      initially Ø.
invocations can start once the node has received the snapshot                    Completed-ops, queue of operations that have been simulated,
and has the active status. A precondition for performing most                    initially Ø.
of these actions that the node is in the relevant focal point.                   Fig. 6 FPE server signature and state for node i and object obj of variable type
This property is covered in most cases by the integrity                          τ = <V, v0, invocations, responses, δ >
requirements of the LBcast and Geocast services, which imply
                                                                                 Signature:
that these actions only happen when the node is in the
appropriate focal point[2] [3].                                                  Input:
Signature:                                                                       Invoke (inv) obj, p, inv ∈ invocations, P ∈ Q
Input                                                                            Geocast –rcv (<response, resp, oid, loc>)obj,i , resp ∈
                                                                                 responses, oid ∈ U
Geocast-rcv (< invoke, inv, oid, Loc>) obj,i ,inv ∈ invocations,
                                                                                 Geo-update (l,t)obj,i, l ∈ L , t E R > 0
oid ∈ U,loc ∈ L    (i.e. oid: object identifier , loc: location)
                                                                                 Output:
Lbcast-rcv (<Join-req, jid >)       obj,i   , Jid   ∈ T (i.e. jid: join
identifier)                                                                      Geocast (m) obj,i, m ∈ invoke×invocations×U ×L×L
                                       ∈ T, v ∈ V
Lbcast-rcv (<Join-ack, jid, v>) obj,I, Jid                                       respond (resp) obj,p, resp ∈ responses, p ∈ Q
                                                                                 State:
Lbcast-rcv (<invoke, inv, oid, loc>) obj, i, inv ∈ invocations,
oid ∈ U, loc ∈ L                                                                 Fp-location ∈ 2L, a constant, the locations of the focal point
Geo-update (l,t)obj,I ,l ∈ L ,t ∈ R>0                                            Clock ∈ R ≥ 0, the current time, initially 0, updated by
                                                                                 geosensor
Output:
                                                                                 Location ∈ L, the current physical location, updated by
Geocast (<response, resp, oid, loc>)          obj,i,   resp   ∈ Responses,       geosensor
oid ∈ U , loc ∈ L
                                                                                 ready-responses C Q×responses, a set of operation responses,
Lbcast (< Join-req, Jid>) obj, i, jid ∈T                                         initially Ø
Lbcast (< Join-ack, Jid, v>) obj, i, jid ∈ T, v ∈ V                              Geocast-queue, a queue of messages to be geocast
Lbcast (< invoke, inv, oid, loc>) obj, i, inv ∈ invocations, oid                 ongoing-oids C U, a set of operation identifiers, initially Ø
∈ U, loc ∈ L                                                                     Transitions:
Internal:                                                                        Input invoke (inv)obj, P
Join ( ) obj,i                                                                   Effect:
Leave ( ) obj,i                                                                  New-oid <clock, p>
Simulate-op(inv)obj,i , inv ∈ invocations.                                       Enqueue (geocast-queue, < invoke, inv, new-oid, location, fp-
State:                                                                           location>)
Fp-location ∈ 2L, constant, locations defining the focal point                   Ongoing-oids ongoing-oids U {new-oid}
under consideration                                                              Input geocast-rcv (<response, resp, oid, loc>)obj,i
Clock    ∈ R>0, the current time, initially                                      Effect:
0, updated by the geosensor.                                                     If (oid ∈ ongoing –oids) then
Location ∈ , L the current physical location, updater by the                     <C, p>          oid
geosensor                                                                        ready-responses ready-responses U {<p, resp>}
Status ∈ {idle, joining, listening, active}, initially active if                 Ongoing-oids Ongoing-oids \ {oid}
node is in FP-location and idle otherwise.                                       Input geo –update (l,t)obj,i
Join- id    ∈ T, unique id for current join request, initially <0,               Effect:
i0>.                                                                             Location L
Lbcast- queue, a queue of messages to be sent by the LBcast,                     Clock ← t
initially Ø.
                                                                                 output geocast (m) obj,i
Geocast-queue, a queue of messages to be sent by the
                                                                                 Precondition:
Geocast, initially Ø.




                                                                             6
                                                                           206                                   http://sites.google.com/site/ijcsis/
                                                                                                                 ISSN 1947-5500
                                                                       (IJCSIS) International Journal of Computer Science and Information Security,
                                                                       Vol. 8, No. 2, May 2010

                                                                                       Input LBcast –rcv (< invoke, inv, oid, loc>) obj,i
Peek (geocast-queue) =m                                                                Effect:
                                                                                       If ((status=listening V active) ^
Effect:
                                                                                        (<inv, oid, loc> ∉ pending-ops U completed-ops)) Then
Dequeue (geocast-queue)                                                                Enqueue (pending-ops,<inv, oid, loc>)
Output respond (resp) obj,i                                                            Internal simulate-op (inv) obj,i
Pre conciliation:                                                                      Precondition:
                                                                                       Status=active
< P, resp> ∈ ready-responses
                                                                                       Peek (pending-ops) =<inv, oid, loc>
Effect:                                                                                Effect:
Ready-responses ← ready-responses \{<p, resp >}                                        (Val, resp) δ (inv, val)
Fig. 7 FPE client for client i and object obj of variable type   τ   = <V, v0,          Enqueue (geocost- queue,< response, resp, oid, loc>)
invocations, responses, δ >                                                            Enqueue (completed-ops, Dequeue (pending-ops))
                                                                                       Internal leave ( )obj,i
Focal Point Emulator Server Transitions                                                Precondition:
Internal join ( )obj,i                                                                 Location ∉ fp-location
Precondition:                                                                          Status ≠ idle
Location ∈ fp-location                                                                 Effect:
Status=idle                                                                            Status idle
Effect:                                                                                Join-id <0,i0>
Join-id ←<clock, i>                                                                    Val v0
Status← joining                                                                        answered -join- reqs← Ø
Enqueue (lbcast-queue, <join-req, join-id>)                                            Pending –ops ← Ø
Input lbcast-rcv (< join-req, jid>) obj,i                                              Completed-ops ← Ø
Effect:                                                                                Lbcast-queue ← Ø
If ((status=joining)) ^ (jid=Join-id)) then                                            Geocast-queue ← Ø
Status listening                                                                       Output Lbcast (m) obj,i
If ((status=active))) ^ jid ∉ answered-join-reqs)) then                                Precondition:
Enqueue (LBcast-queue,< join-ack, jid, val>)                                           Peek (Lbcast-queue) =m
Input Lbcast-rcv (<join-ack, jid, v>)obj,i                                             Effect:
Effect:                                                                                Duqueue (Lbcast- queue)
Answered-join-reqs answered-join-reqs U {jid}                                          Output geocast (m) obj,i
if ((status=listening) ^ (jid =join-id)) then                                          Precondition:
Status active                                                                          Peek (geocast-queue) =m
Val v                                                                                  Effect:
Input geocast –rcv (< invoke, inv, oid, loc, fp-loc>)obj,i                             Dequeue (geocost- queue)
Effect:                                                                                Input get-update (l,t) obj,i
If (fp-loc=fp-location) then                                                           Effect:
If (<inv, oid, loc> ∉ pending-ops U completed ops) then                                Location l
Enqueue (Lbcoast-queue, <invoke, inv, oid, loc>)                                       Clock t

                         Fig. 8 FPE server transitions for client i and object obj of variable type   τ   = <V, v0, invocations, responses, δ >.
               III. A SPECIFICATION IN TUG
TUG specification language consists of 3 parts: a name part                            taken that must occur in the input data. A terminal node can
where the title with input/ output parameters is given, on                             be a literal which is any string enclosed in a pair of quotes. A
analysis part where the input data is defined, and an anatomy                          constraint wrapped in braces places the conditions such as
part where the output data is generated. The name part                                 type checking indicates a taken that must occur in the input
contains a module or schema title with input/ output                                   data. A terminal node can be a literal which is any string
parameters are enclosed in parentheses. The analysis part                              enclosed in a pair of quotes. A constraint wrapped in braces
contains the rules for analyzing the input data. To analyze the                        places the conditions such as type checking on a terminal
input data, Definite Clause Grammars (DCGS) are used to                                node. Table I includes all operators used in the conditions. An
represent the rules to perform the syntax analysis. Each rule of                       input is parsed into a tree representation that takes the form of
a DCG expresses a possible form for a non terminal, as a                               a prolog list with a node name acting as the relationship
sequence of terminals with optional constraints on the                                 symbol of the input data [4]. This tree representation will be
terminals and non terminals. Non terminal nodes in uppercase                           the input to the anatomy analysis part of the TUG
indicate constituents. A terminal node in lowercase indicates a                        specification.




                                                                                   7
                                                                                  207                                    http://sites.google.com/site/ijcsis/
                                                                                                                         ISSN 1947-5500
                                                                  (IJCSIS) International Journal of Computer Science and Information Security,
                                                                  Vol. 8, No. 2, May 2010



                                                                                                            First _ name
                                                                                                       {String (first_ name)}
          Operators                            Description                        Indicates that NAME is a concatenation of last_ name and
                                                                                  first_ name. The last_name and first_name must be of string
            any (t)                        t belongs to any type
                                                                                  type.
            bool (t)                          t is a Boolean                      The kleene closure notation (*) means zero or more element,
          character (t)                       t is a character                    over the node. Thus,
            digit (t)                            t is a digit                                                    MAIL- Box*
        equal_ to (t1, t2)                    t1 is equal to t2                                                       Mail
            float (t)                            t is a float
                                                                                                                 {Text (mail)}
      greater_ than (t1, t2)                t1 is greater than t2
                                                                                  Indicates that NAIL- Box contains zero or more mails, each of
           integer (t)                         t is an integer
                                                                                  which is a text.
           length (t)                       length of a string t
                                                                                  The positive closure notation (+) means one or more elements
        less_than (t1, t2)                    t1 is less than t2                  over the node. So, that:
    lowercase_charater (t)               t is a lowercase character                                               MAIL+
        member (t1, t2)                     t1 is a member of t2                                                 Message *
                                                                                                                  Letters
             not (t)                       logical negation of t
                                                                                                             {Letter (letters)}
       remainder (t1, t2)                 remainder for integer
            string (t)                       division t1 / t2                     Indicates that MAIL contains at least one message, which may
                                                t is a string                     contain zero or more letters in it. If a message contains no
             text (t)                                                             letters, then the MAIL is an empty Mail
    uppercase_ character (t)                     t is a text
                                      t is an uppercase character                 IV. RAPID PROTOTYPING PROCESS VIA SOFTWARE
            word (t)                                                              TRANSFORMATIONS
                                                t is a word
                                                                                       The prototype serves as a basis for discussion to help the
                                                                                  specifier and the user to read just the user requirements and
         TABLE I OPERATORS USED IN THE CONDITIONS
                                                                                  specifications. Feedback from the user is obtained to decide
                                                                                  whether the change is minor or major. If the change is minor,
                                                                                  A Change Request Script (CRS) specifying the change is
The specification is structured with regular expression                           written to update the specification and the prototype. If a
notations (Union, Positive Closure, and Concatenation). Each                      major change is needed, the specifier may rewrite the
non terminal node structures its tree according to one of the                     specification and rederive a new prototype from the start. A
regular expression notations. Adding tree structures to a                         major change may involve the structure of the specification to
specification allows a software developer to construct the                        be modified. This prototyping process continues until the
specification in a structured way for dealing with complexity                     requirements have been thoroughly exercised and the user is
[4].                                                                              satisfied with the demonstrated behavior of the prototype. The
                                                                                  results of the prototype evolution are a set of modular TUG
                                                                                  specifications for the proposed system. In addition, a set of
The union notation is indicated by a vertical bar sign ( | )                      CRSs record the design decisions made during the
suffixed to the node name, so that                                                transformations [4].There is no existing specification language
                                FLAG |                                            that can support abstractions for all features of Software
                                                                                  systems. Therefore, a specification language must make
                                 "On"                                             developers to easily extend and modify prototypes. To support
                                                                                  easy modifications to the prototype, the TUG specification
                                 "Off"
                                                                                  language was designed to support the construction of a
Indicates that FLAG is one of the alternatives "on", and" off".                   specification in a structured manner with regular expression
                                                                                  notations. The modules in a derived prototype from the
The concatenation notation is indicated by an ampersand sign
                                                                                  specification can be easily located, modified, or extended in
(&) suffixed to the node name, so that
                                                                                  terms of these regular expression notations. The rapid
                               NAME &                                             prototyping approach using TUG can be incorporated into any
                                                                                  Software development process. It is intended that each
                               Last_name                                          evolution of the specification that is synthesized by the
                          {String (last_name)}                                    specifier should be formally recorded using TUG, and that the




                                                                              8
                                                                             208                             http://sites.google.com/site/ijcsis/
                                                                                                             ISSN 1947-5500
                                                      (IJCSIS) International Journal of Computer Science and Information Security,
                                                      Vol. 8, No. 2, May 2010



prototype derived from the specification should be exercised                                        …………..
with the users participating in the user requirements analysis                                            βn
process. The specification can then be reasoned with and                                              <{Φn}>
expected behavior can be validated. The benefits of rapid                                         α β1< { Φ1}>
prototyping have been identified to include [5]:-                                                 α β2< { Φ2}>
  • Rapid prototyping is available in the front end of the                                          ……………
      Software life cycle to allow early detection of errors.                                       ……………
  • Unclear and imprecise user requirements can be clarified                                      α βn< { Φn}>
      by rapid prototyping.                                                                 In rule1, a union nonterminal node α in the
  • Execution of the prototype supplements inspection and               analysis tree indicates that α is one of the alternatives, β1,
      formal reasoning as means of analysis of the                      β2,….., βn. If βi is a literal, there is no Φi associated with βi.
      specification.                                                    Each translated DCG represents an alternative [9] [10].
  • The underlying theory of the TUG specification language                                           &- def 2
      is DCGs, which can be executed directly in the prolog                                              α&
      environment .There is a close correspondence between                             β1     where α is a nonterminal node
      TUG and prolog, which makes the process of                                          < {Φ1}> βi is a nonterminal
      transformation relatively mechanical. In this approach,                      or terminal node with condition tests Φ i C
      DCGs are used as an intermediate form for aiding the                                                β2
      transformation process. Although DCGs are syntactic for                                         <{Φ2}>
      prolog, a prototype in DCGs seams difficult to read,                                               -----
      understand, and maintain.                                                                          -----
  • Whenever there is a change in the user requirements,                                                  βn
      there many are no need to rederive the prototype from                                           <{ Φn}>
      scratch if the change is trivial. ACRS is written to update                α β1<{ Φ1}> β2<{ Φ2}> …….. βn <{ Φn}>
      the prototype only in response to the revised specification                 In rule 2, α a concatenation nonterminal node α in the
      .A redervation of prototype in prolog from the start is           analysis tree indicates that & is a concatenation of β1, β2,…..,
      avoided is modified [6] [7].                                      β n. If βi is a literal, there is no Φi associated with βi. Each
  • The rapid prototyping approach supports formal                      translated DCG represents a concatenation form
      requirement specifications written in TUG.                                                       *def 3
                                                                                                          α*
  • The prototype is exercised to demonstrate the system
      behavior in the prolog environment. A driver that reads in                       β1     where α is a nonterminal node
                                                                                         < {Φ1}>        β1 is a nonterminal
      the input data and then calls the main program with
                                                                                  or terminal node with condition tests Φ i C
      parameters needs to be constructed manually. The set of
                                                                                                          β2
      transformation rules are given below. The conventions
                                                                                                      < { Φ2}>
      are:
  - C: is a finite set of condition tests and has the for                                              -------
                                                                                                       -------
  {C1, C2,….,Cn} with n ≥ 1 where ci is a TUG condition test;
                                                                                                          βn
 - Y is a finite set of dummy non terminal or terminal node;
                                                                                                      <{Φn}>
 - Names of predicates in prolog are in all lowers case letters.
                                                                                                      α [ ]
 - Names of variables in prolog are in all upper case letters.                α β1<{ Φ1}> β2<{ Φ2}> . .…. .…….. βn <{ Φn}> α
 - Q is a finite set of prolog procedure calls and has the form            In rule 3, a kleene closure nonterminal node α in the
    {q1, q2…q n} with n ≥ 1 where qi is a prolog predicate for          analysis tree indicates that α is a sequence of zero or more
  which a Prolog definition has been given, and. < > encloses           occurrence of β1, β2,….., βn. If βi is a literal, there is no Φ
  optional syntactic Items [8].                                         associated with βi two translated DCGs represent a kleene
  ** The following four rules translate the analysis part of the        closure form [5][11].
  TUG specification into DCGs. Each non-terminal node in the                                          + - def 4
  analysis tree structures its subtrees according to one of the                                          α+
  structure notations each structuring operation can be                             β1           where α is anonterminal node
  transformed into a DCG form by applying the following four              <{ Φ1}> β i anonterminal or terminal node with condition
  rules in straight forward manners[ 5][9].                                                          tests Φi C
                                |-def 1                                                                   β2
                                  α1                                                                  < { Φ2}>
                  β1 where α is a nonterminal node                                                      ------
                               < {Φ1}>                                                                  ------
                                   β2                                                                     βn
                              < { Φ2}>
                             …………..




                                                                    9
                                                                 209                               http://sites.google.com/site/ijcsis/
                                                                                                   ISSN 1947-5500
                                                       (IJCSIS) International Journal of Computer Science and Information Security,
                                                       Vol. 8, No. 2, May 2010



                            <{Φn}>                                   associated with βi.. Two translated DCGs represent a positive
        α β1<{ Φ1}> β2<{ Φ2}> …….. βn <{ Φn}>                        closure form [5][12].
       α β1<{ Φ1}> β2<{ Φ2}> …….. βn <{ Φn}> α                       To demonstrate the use of transformation rules presented in
In rule 4, α a positive closure nonterminal node α in the            this section, the application of Rules 1-4 to the analysis tree of
analysis tree indicates that α is a sequence of one or more          the related work approach specification produces the
occurrences of β1, β2... βn. If βi is a literal, there is no Φi      following results
  (1) SEQUENCE→UNSORTED- IDS                                                                           α&
  (2) SEQUENCE →SORTED- IDS                                                          β 1 where α is a nonterminal node
  (3) UNSORTED→ List 1 +X{integer(X)}                                           β2      β i is a nonterminal node or statement
                       List 2+Y {integer(y)},                                                        ………
                       Greater-than(X, Y)}                                                                βn
                       Rest_of_elements                                ε:-Y1<{Ψ1}> Y2<{Ψ2}>…… Yn<{ Ψn}> Where Ψ1 C Y
  (4) SORTED →ASCENDING-SEQUENCE                                                             ε is uppercase using α
  (5) ASCENDING_SEQUENCE→ Element {integer                                  Yi is uppercase using β i if β i is a nontreminal node;
       (element)}                                                                                 Otherwise Yi = β i
                                                                     In Rule 6, a concatenation nonterminal node α in the anatomy
  (6) ASEDING_SEQENCE→ Element {integer (element)}                   tree indicates that α is a concatenation of β1, β2,…, β n. The
            ASCENDING-SEQUENCE                                       translated DCG represents a concatenation form.
                                                                                                     * -def 7
The following four rules translate the anatomy tree of the                                             α*
specification into DCGs. The rules are similar to the rules for                      β 1 where α is a nonterminal node
translating the analysis tree. The difference is that we use               β 2 β i is a nonterminal node ,terminal node, or statement
the":- "symbol instead of symbol " ". The use of the " "                                            ……….
symbol in the rules for the analysis part pf a TUG                                                        βn
specification denotes a derivation of a tree, an involvement                                         ε:- [ ]
of pattern matching, and an unification of variables with the          ε:- Y1<{Ψ1}> Y2{Ψ2}>…….Yn<{Ψn}> ε where Ψ1 C Y
input values in prolog and the use of the" :-" symbol in the                                 ε is uppercase using α
rules for the anatomy part of the specification perfumes the                Yi is uppercase using β i if β i is a nontreminal node;
same operations. The uses of the " "and":-" symbols are                                           Otherwise Yi = β i
just for the syntactic purpose. The out puts of the rules for the    In Rule7, a kleene closure nonterminal node α in the anatomy
analysis part of a TUG specification produce a tree unified          tree indicates that α is a sequence of zero or more occurrence
with the input values that is the input to the rules for the         of β1, β2,…, β n . Two translated DCGs represent a kleene
anatomy part of the specification. The rules for the anatomy         closure form.
part of a TUG specification reads in the tree and performs                                           +-def 8
exact unifications on the variables to produce outputs                                                 α+
Another difference is that dummy nodes appear in the rules.                          β 1 where α is a nonterminal node
The reason for having dummy nodes is that often only the                   β 2 β i is a nonterminal node ,terminal node, or statement
parts of tree are referenced in the anatomy tree of the                                               ……
specification. The remaining unreferenced parts of the tree                                             βn
still need to be unified in the course of patter matching.                         ε:- Y1<{Ψ1}> Y2{Ψ2}>…….Yn<{Ψn}>
Dummy nodes are obtained by referring back to the analysis                        ε:- Y1<{Ψ1}> Y2{Ψ2}>…….Yn<{Ψn}> ε
tree of the specification [7] [12].                                                              Where Ψ1 C Y
                                                                                             ε is uppercase using α
                                  | - def 5                                 Yi is uppercase using β i if β i is a nontreminal node;
                                 α|                                                               Otherwise Yi = β i
             β1       where α is a nonterminal node                  In Rule 8, a positive closure nonterminal node α in the
                  β2        β i is a nonterminal node                anatomy tree indicates that α is a sequence of one or more
                                ………..                                occurrences of β1, β2,…, β n .
                                     βn                              Two translated DCG represent a positive closure form. The
                 ε: -Y1 where ε is uppercase using α                 application of Rules 5-8 to the anatomy tree of the problem
                ε: -Y2 where Yi is uppercase using β i               specification produces the following results
                               ………….                                    (7) SEQUENCE:- UNSORTED
                               ε: -Yn                                   (8) SEQENCE:-SORTED
In Rule 5, a union nonterminal node α in the anatomy tree               (9)UNSORTED: -T-L1=Y:: List 2
indicates that α is one of the alternatives, β1, β2,…, β n .Each           T-L2=X:: rest_of_elements
translated DCG represents an alternative.                                  T-L=List 1< >T-L1< >T-L2
                             &-def 6                                       Call IDS_sort (T-L)




                                                                    10
                                                                   210                            http://sites.google.com/site/ijcsis/
                                                                                                  ISSN 1947-5500
                                                   (IJCSIS) International Journal of Computer Science and Information Security,
                                                   Vol. 8, No. 2, May 2010



    (10) SORTED: - ASCENDING_SEQUENCE                                 Output "config"
    (11) ASCENDING_SEQUENCE:-Output element                           Output "reconfig"
    Output ' '
    (12) ASCEDING_SEQENCE:-Output element                             END OF ANATOMY
    Output ' '                                                        END OF MODULE a_ Listing_ of_ transitions
    ASSCENDING_SEQUENCE
                                                                       B. The Application of Transformation Rules to the Above
V. A TUG SPECIFICATION FOR IMPLEMENTING                               Specification Module Results
ATOMIC READ/WRITE SHARED MEMORY IN MOBILE
AD HOC NETWORKS APPLICATION                                           In the Following Prototype in Prolog:
         This section will illustrates the usage of TUG for           Transition_ type (transition_ type
implementing atomic read/ write shared memory in mobile ad            (TRANSITION))
hoc networks. A specification in TUG is formalized
incrementally in a modular and Top-down manner the                    Transition (TRANSITION).
example also illustrates how the language supports module             Transition (Transition ([ ]))
independence via the language patterns. The Geoquorums
approach, for implementing atomic read/ write shared                  [ ].
memory in mobile ad hoc networks. This approach is based an           Transition (transition ("put", "get", "confirm", "config",
associating abstract atomic object, with certain geographic           "reconfig"))
locations. We assume the existence of local points, geographic
areas that are normally "populated" by mobile nodes. The              ["put"],
Geoquorum algorithm uses the fault –prone focal point                 ["get"],
objects to implement atomic read/write operations on fault –
tolerant virtual shared object. Te Geoquorums algorithm uses          ["confirm'],
a quorum- based strategy in which each quorum consists of a           ["config"],
set of focal point objects. The quorums are used to maintain
the consistency of the shared memory and to tolerate limited          ["reconfig"]
failures of the focal point objects which may be caused by            Transition (TRANSITION).
depopulation of the corresponding geographic areas .Overall,
the new geoquorums algorithm efficiently implements read              Transition_ type (transition_ type
and write operations in a highly dynamic, Mobil network.
                                                                      ((TRANSITION)):-
A. A First Attempt at the Specification
                                                                      Transition (REANSITION).
MODULE a_ listing_ of _transitions
                                                                      Transition (Transition ([ ])).
(in: TRANSITION_TYPE)
                                                                      Transition (Transition("put", "get", "confirm", "config",
ANALYSIS                                                              "reconfig")) :-
TRANSITION _TYPE &
                                                                      n1,
TRANSITIONS *
                                                                      Write ("put", "get", "confirm", "config", "reconfig")),
"Put"
                                                                      Transition (TRANSITION).
"get"
"confirm"                                                             (2)The Following CRS is Further Refinement on Each
                                                                      Invocation
"config"
                                                                      Replace TRANSITION* under TRANSITION _Type &
"reconfig"
END OF ANALYSIS"                                                      With
ANATOMY                                                               VARIABLE_TYPE_TRANSITION.
Transition_ type&                                                     "put_ invocation"
Transition*                                                           "get_ invocation"
Output n1                                                             "confirm_ invocation"
Output "put"
                                                                      "config _invocation"
Output "get"
                                                                      "reconfig_ invocation"
Output "confirm"
                                                                      Replace Transition* under transition_ type &




                                                                 11
                                                              211                               http://sites.google.com/site/ijcsis/
                                                                                                ISSN 1947-5500
                                                      (IJCSIS) International Journal of Computer Science and Information Security,
                                                      Vol. 8, No. 2, May 2010



With                                                                Variable_type_transition(get_invocation
                                                                    (GET_INVOCATION)) →
Variable_Type_Transition |
                                                                    get_invocation (GET_INVOCATION).
Variable Transition &
                                                                    put_invocation('put_invocation_separator,
Output n1
                                                                    PUT_INVOCATION_SECTION)) →
Output "put _ invocation"
                                                                    ['put_invocation_separator'],
Output" get _ invocation"
                                                                      put_invocation_section (PUT_INVOCATION_SECTION).
Output "confirm – invocation"
                                                                      put_invocation_section ([ ])).
Output "config – invocation"
                                                                    put_invocation_section(put_invocation_section(PUT_INVOC
Output" reconfig – invocation"                                      ATION,'put_invocation_separator',
                                                                    PUT_INVOCATION_SECTION)) →
C. A CRS for This Refinement is shown below
                                                                    [PUT_INVOCATION],
Replace PUT_INVOCATION_SECTIONS* under PUT_
INVOCATION                                                          {put_ack_response        (PUT_INVOCATIO,              NEW_VALUE,
                                                                    NEW_TAG,
With
                                                                      NEW_CONFEG_ID),
Put_ invocation
                                                                    Length (new_config_id >config_id)
{new_value (put_invocation)
                                                                    [put_invocation→ PUT_ACK_RESPONSE],
New_tag (put_ invocation)
                                                                    Put_invocation_section (PUT_INVOCATION_SECTION).
New_config_id (put_ invocation)}
                                                                    Transition_type (transition_type (TRANSITION)): -
"get- invocation".
                                                                    transition (TRANSITION)
ReplaceGET_INVOCATION_SECTIONS*underGET_
INVOCATION & with                                                   transition (transition ([ ])).
get_invocation                                                      Transition(transition(PUT_INVOCATION,
                                                                    GET_INVOCATION,CONFIG_INVOCATION,
{new_config_id (get_invocation)}                                    RECONFIG_DONE _INVOCATION)): -
"get_invocation"
                                                                    get_invocation (GET_INVOCATION),
 At this stage, the application of transformation rules to the      transition (TRANSITION).
above two CRSs result in the following Prolog to update the
prototype.                                                          Get_invocation (get_invocation (GET_INVOCATION)):-
Transition_Type (transition_type (TRANSITION)) →                    get_invocation (GET_INVOCATION).
transition (TRANSITION).                                            confirm_invocation        (confirm_      invocation       (CONFIRM-
                                                                    INVOCATION)):
                                                                    confirm_ invocation (CONFIRM_ INVOCATION).
transition (transition ([ ])) → [ ]
                                                                    get_invocation(get_invocation
transition(transition(PUT_INVOCATION,GET_INVOCATI                   ('initial_get_invocation_separator',
ON,
CONFIRM_INVOCATION,CONFIG_INVOCATION,RECO                           GET_INVOCATION_SECTION)):-
NFIG_INVOCATION)) →
                                                                    write ('get_invocation'),
put_invocation (PUT_INVOCATION),
                                                                    write ('put_invocation'),
(GET_INVOCATION).
                                                                    write ('config_invocation'),
transition (TRANSITION)
                                                                    write ('reconfig _done_invocation'),
Variable_type_transition(put_invocation
(PUT_INVOCATION)) →                                                 confirm_invocation(confirm_invocation('new_tag',
put_invocation (PUT_INVOCATION).
                                                                    CONFIRM_INVOCATION_SECTION)): -




                                                                 12
                                                                 212                             http://sites.google.com/site/ijcsis/
                                                                                                 ISSN 1947-5500
                                        (IJCSIS) International Journal of Computer Science and Information Security,
                                        Vol. 8, No. 2, May 2010



nl,                                                     output 'get_invocation'
write ('put_invocation'),                               output 'get _ ack _ response'
write ('get_invocation'),                               output ' . '
                                                        output nl
write ('config_invocation'),                            config_invocation &
write ('recon_done_invocation'),                        output ' config_invocation'
D. The Further Refinement                               output ' config _ ack _ response'
Replace 'put_invocation' under                          output ' , '
NEW_CONFIG_ID & with                                    output ' recon_done _ invocation'
PUT_ACK_RESPONSE                                        output ' recon_done _ ack'
Stop &                                                  output ' . '
'Stop'                                                  output n1
'     '                                                 After a successive of refinements to the original specification,
                                                        the final complete specification for implementing atomic
replace 'get_invocation' under
                                                        read/write shared memory in mobile ad hoc networks is
NEW_CONFIG_ID & with                                    shown below.
GET_ACK_RESPONSE                                        MODULE a _ Listing _ of _ Transitions
Stop&                                                   (in: TRANSITION _ TYPE)
'Stop'                                                  ANALYSIS
'     '                                                 TRANSITION _ TYPE &
replace 'config_invocation' under                       TRANSITION*
NEW_TAG & with                                          PUT_INVOCATION |
CONFIG_ACK_RESPONSE                                     NEW_CONFIG_ID&
Stop&                                                   PUT_ACK_RESPONSE
'Stop'                                                  Stop&
'     '                                                 'Stop'
replace 'recon_done_invocation' under                   Get_ INVOCATION |
NEW_CONFIG_ID & with                                    NEW_CONFIG_ID&
RECON_DONE _ ACK                                        GET_ACK_RESPONSE
Stop&                                                   Stop&
'Stop'                                                  'Stop'
'     '                                                 ' '
replace transition_ type & with                         CONFIG_INVOCATION |
output 'transition Analysis'                            NEW_TAG&
output nl                                               CONFIG_ACK_RESPONSE
transition*                                             Stop&
output 'transition:'                                    'Stop'
output 'put_invocation'                                  ' '
output 'put _ ack _ response'                           RECONFIG_DONE_ INVOCATION |




                                                   13
                                                   213                             http://sites.google.com/site/ijcsis/
                                                                                   ISSN 1947-5500
                                           (IJCSIS) International Journal of Computer Science and Information Security,
                                           Vol. 8, No. 2, May 2010



NEW_CONFIG_ID&                                                     VI. CONCLUSIONS AND FUTURE WORK
RECON_DONE_ACK                                                 An approach was developed to support rapid prototyping
                                                           via software transformations by deriving a prototype in prolog
Stop&
                                                           from a specification in TUG. We didn't directly use the prolog
'Stop'                                                     language to specify user requirements, programming
                                                           languages more or less concentrate on how rather than what
' '                                                        and are generally unsuitable for specification purposes. In
END OF ANALYSIS                                            addition, in a specification in prolog lacks modularity in
                                                           contrast to a specification in TUG. Since the main purpose of
ANATOMY                                                    a specification is to aid the understanding of the user
transition_type &                                          requirements, it is useful if a specification can be read and
                                                           understood. Modularity helps specifiers to read and maintain
output' Transition Analysis'                               in a manageable way. TUG provides modularity to help
output nl                                                  specifiers to specify a system in a hierarchical manner. A set
                                                           of modules are specified and then composed into a system.
transition*                                                The system is tested in pieces corresponding to the modular
output' transition:'                                       specification .In contrast to a specification in TUG, a
                                                           specification in prolog is relatively difficult to maintain. Rapid
put _invocation |                                          prototyping via software transformations helps to build
new_config_id &                                            prototypes automatically from specifications. In this paper, a
                                                           formal method with TUG was presented to support the rapid
output 'put_ invocation'                                   prototyping via software transformations process in which a
output 'put_ack_response'                                  prototype can be built quickly and cheaply. Automation of the
                                                           application of software transformations reduces the labor
output '. '                                                intensity of developing prototypes manually. Rapid
output nl                                                  prototyping via software transformations also provides
                                                           support for prototype modifications. The rapid prototyping
get _ invocation |                                         approach supports prototype evolution by avoiding complete
new_config_id &                                            retransformation of the prototype from the start whenever
                                                           there is a change made to the specification. To avoid complete
output 'get_ invocation'                                   retransformation, a CRS is written to update the prototype
                                                           only in response to the minor changes to the specification,
output 'get_ack_response'
                                                           involving the nodes to be modified, extended, relaxed, or
output ', '                                                refined. If a major change is needed, the specification may
                                                           need to be rewritten and a new prototype may be derived from
config- invocation |
                                                           the start. A major change may involve the structure of the
new-tag &                                                  specification to be modified. Like other formal specification
                                                           languages, the TUG specification language may not provide
output 'config- invocation'
                                                           enough abstractions for modeling some properties of software
output 'config -ack-response'                              systems such as non- functional properties. Therefore, the
                                                           geoquorum approach for implementing atomic read/write
output ' , '                                               shared memory in mobile ad hoc networks encourages
recon - done- invocation |                                 specifiers to manually add additional code to the derived
                                                           prototype for demonstrating such kind of properties of
new-config –id &                                           systems. The rapid prototype approach supports the quick
output ' reconfig - done- invocation'                      construction of a prototype with a high degree of module
                                                           independence. Module independence has a particular
output ' recon - done- ack'                                importance in this approach because of the need for
                                                           modifications to the prototype. It remains an open question to
output '.'                                                 determine how to choose a good set of focal points, how to
output nl                                                  construct a map of focal points in a distributed fashion, and
                                                           how to moodily the set of focal points dynamically. Overall,
END OF ANATOMY;                                            the FPO Model will significantly simplify the development of
                                                           algorithms for mobile, in highly dynamic networks. Finally,
                                                           there exist many techniques to do the phases of software
END OF MOUDULE a-Listing-Of-Transitions.
                                                           lifecycle for any application.




                                                      14
                                                      214                             http://sites.google.com/site/ijcsis/
                                                                                      ISSN 1947-5500
                                                  (IJCSIS) International Journal of Computer Science and Information Security,
                                                  Vol. 8, No. 2, May 2010



                  ACKNOWLEDGMENT                                    Annual IEEE Symposium on Logic in Computer Science
                                                                    (LICS,00),JUNE(2000),26-29,California ,2000,PP:242-249.
    The authors would like to thank the INFOS 2010                  [5] Chia- Chu Chiang," Automated Rapid Prototyping Of Tug
Conference (Cairo University)-EGYPT reviewers for learning          Specifications Using Prolog ", Proceedings Of: Information
from their constructive comments and suggestions in                 And Software Technology 46(2004), PP: 857-873.
preparing the scientific papers.                                    [6] Chia-Chu Chiang," Automated Rapid Prototyping of TUG
                                                                    Specifications Using Prolog", Proceedings of: Information
                                                                    and Software Technology 46(2004), PP: 857-873.
                      REFERANCES                                    [7] O.J.Dahl, O. Owe, Formal Methods and the RMODP,
                                                                    Research Report No.261, Department of Information,
[1] C.Chiang,J.E.Urban, "Validating Software Specification          University of Oslo, Norway, 1998.
against User Claims", Proceedings of the Twenty-Third               [8] A. Evans, UML Class Diagrams –Filling the Semantic
Annual International Computer Software and Applications             Gap, Technical Report, York University, 1998.
Conference (COMPSAC 2000),2000,PP:104-109.                          [9] IEEE, IEEE standard for a High Performance Serial Bus,
[2] DOLEV, S., Gilbert, S.LYNCH, N.A., SHVARTSMAN,                  Standard 1394, August 1995.
A.A., Welch, J.L.: " Geoquorums: Implementing Atomic                [10] M.Liu Yanguo, Proof Patterns for UMI-Based
Memory in Mobile Ad Hoc Networks". In: Proceeding of the            Verification, Master Thesis ECE Department, University of
17th International Conference on Distributed Computing, PP.         Victoria, Victoria, Canada, October 2002.
306-320 (2003).                                                     [11] I.Traore, D.Aredo, H.Ye," An Integrated Framework for
[3] Haas, Z.J., Liang, B.: "Ad Hoc Mobile Management with           Formal Development of Open Distributed Systems," In:
Uniform Quorum Systems". IEEE/ACM Transactions on                   Information and Software Technology 46 (2004) 281-286.
Networking 7(2), PP: 228-240 (2000).                                [12] I. El-Far, Automated Construction of Software Behavior
[4] B.CMoszkowski,"A Complete Axiomatization Of Interval            Models, Master's Thesis, Florida Institute of Technology
Temporal Logic With Infinite Time ", Proceedings of the 15th        Melbourne, Fl, 1999.




                                                               15
                                                             215                             http://sites.google.com/site/ijcsis/
                                                                                             ISSN 1947-5500
(IJCSIS) International Journal of Computer Science and Information Security,
Vol. 8, No. 2, May 2010




           16
           216                             http://sites.google.com/site/ijcsis/
                                           ISSN 1947-5500