Outline Dynamic Adaptation of Mobile Ad Hoc Routing Protocols

Outline Dynamic Adaptation of Mobile Ad Hoc Routing Protocols Hwee Xian TAN1,2 1Department Overview Routing Protocols for MANETs Related Work and Motivation Our adaptive solution Dynamic Adaptation of Frequency of Broadcasts Limiting Control Overheads Using Link Stability Dynamic Topology Control to Reduce Interference of Computer Science School of Computing National University of Singapore 2Institute for Infocomm Research Agency for Science Technology and Research Conclusion & Future Work Overview MANETs – Mobile Ad Hoc Networks No existing infrastructure No central administration Mobile nodes Useful in hostile terrains, military and civilian applications Challenges of MANETs Network dynamics Network topology Data congestion Shared medium contention Traffic loads Traffic patterns Lack of infrastructure and central control Resource constraints Routing Protocols for MANETs Proactive routing protocols OLSR – Optimized Link State Routing TBRPF – Topology Dissemination Based on Reverse Path Forwarding Limitations of existing protocols Unable to adapt according to network dynamics Preset system parameters may not be suitable for all network conditions Performance degradation when routing protocols are used in environments that they were not designed for Reactive routing protocols DSR – Dynamic Source Routing AODV – Ad Hoc On Demand Distance Vector Hybrid routing protocols ZRP – Zone Routing Protocol CBRP – Cluster Based Routing Protocol Adaptive Protocols Aim to overcome the shortcomings of existing protocols Adapt to network characteristics to improve network performance Some examples include: ADV – Adaptive Distance Vector ARM – Adapting to Route Demand and Mobility SHARP – SHARP Hybrid Adaptive Routing Protocol ASAP – Adaptive Reservation and Pre-allocation Protocol Metrics for Consideration Node mobility Link stability Network size/density Traffic load Traffic pattern Power Directionality of links AODV Routing Protocol Ad Hoc On Demand Distance Vector Dynamic routing protocol Best effort delivery Uses sequence numbers to avoid routing loops Responds quickly to changes in network topology Uses 3 types of control messages: RREQ (Route Requests) RREP (Route Replies) RERR (Route Errors) Dynamic Adaptation of Frequency of Broadcasts Can be incorporated into any reactive protocol that uses periodic beacons to transmit local connectivity information Hello messages used in AODV Problems associated with periodic beacons: Routing overhead Contention of bandwidth with data packets Increased collisions Lower packet delivery ratio Hello messages Used to provide local connectivity information RREP with TTL = 1 Dynamic Adaptation of Frequency of Broadcasts Vary the frequency of periodic broadcasts based on node mobility Node mobility – defined as the rate of change of neighbours Newx = number of new neighbours Left = number of neighbours that have moved away Node mobility, Nm = Newx + Left Dynamic Adaptation of Frequency of Broadcasts IF (Nm < LOW_THRESH) Deviation_Frac = LOW_FRAC; ELSE IF (Nm > HIGH_THRESH) Deviation_Frac = HIGH_FRAC; ELSE Deviation_Frac = 1; Timer for next Hello message = HELLO_INTERVAL * DEVIATION_FRAC Link Stability Measure of the reliability of link between any 2 arbitrary nodes Depends on: Distance between nodes Signal strength of transmitting node Sensitivity of receiving node Antenna gains of transmitter & receiver Environmental conditions Link Stability Unstable networks lead to frequent link breakages More control packets Increased contention of bandwidth and collisions Lowered packet delivery ratio Higher latency Essential for the formation of stable networks Higher reliability Better QoS support Link Stability Relative signal strength between 2 nodes, Relss[x] = 10 lg(Px/Px’) Px = signal strength received in current time period Px’ = signal strength received in previous time period L-REQ Limited Forwarding of RREQ packets to highly mobile nodes C A Source node F B E D Relss[x] < Qthresh (RREQ are discarded) Relss[x] ≥ Qthresh (RREQ are broadcasted) L-REP Limited Initiation of RREP packets by intermediate nodes Intermediate node with valid route, Relss[x] x] < Pthresh Limiting Control Overheads Using Link Stability FOR each RREQ received IF (RREQ is a new request) { IF (receiving node = destination node) Send unicast RREP ELSE // intermediate node IF intermediate node has fresh route to dest L-REP ELSE L-REQ } ELSE Discard RREQ C A B D Intermediate node with valid route, Relss[x] ≥ Pthresh E F RREP packet (unicasted) RREQ packet (broadcasted) Network Size Defined as the total number of nodes in the network Typical networks are heterogeneously distributed 2 main problems associated with network topology: Sparse networks Dense networks Topology Control Deliberate adjustment of certain system parameters (antenna direction, transmission power) to form a particular network topology to improve performance Commonly used to optimize power consumption in MANETs Critical nodes, C[x] = {set of nodes required to transmit data} Non-critical nodes, N[x] = {set of nodes that are not required to transmit data} All neighbouring nodes, A[x] = C[x] U N[x] C[x] ∩ N[x] = Ø Dynamic Topology Control to Reduce Interference Hidden/Exposed terminal problem Critical Neighbour (CN) scheme Dynamic Topology Control to Reduce Interference Critical Neighbour (CN) scheme Measurement of estimated critical transmission range • Ground Reflection model, Pr = Pt× ht2× hr2× Gt× Gr / d4 • Estimated distance between node x and its ith critical neighbour, Edist[xi] = 4√(Pt× ht2× hr2× Gt× Gr / Pr) • Critical transmission range, Ctxn[x] = Max(Edist[xi], Edist[xi+1], …, Edist[xn]), 1 ≤ i ≤ n Estimation of ideal power Taking Gt= Gr = 0 dBm = 1, • Pideal = Pmin_r× Ctxn4× tolerance_factor/ ht2× hr2 Adjustment of ideal power based on constraints Transmission range Interference range Dynamic Topology Control to Reduce Interference /* Handling of Hello packets */ FOR each Hello packet received { IF it is received from a critical neighbour { determine the critical transmission range Ctxn; determine the ideal transmission power Pideal; IF Pideal is within constraints adjust the transmission power accordingly; } // end if update routing table and neighbour tables; } // end for Simulation Environment Network Simulator – GloMoSim MAC protocol – IEEE 802.11 Traffic – CBR (Constant Bit Rate) with packet size of 512 bytes at intervals of 100 ms Terrain size – 2000× 2000 metres Mobility models: Random Waypoint RPGM (Reference Point Group Mobility) Conclusion Small changes to routing protocols to adapt to network characteristics can enable it to improve performance while maintaining interoperability with its unmodified version Future Work Mathematical analysis to determine each of the parameters being used in our simulations Current values are used based on empirical results arising from extensive simulations Improving performance of MANET routing protocols by dynamically adapt them according to other network characteristics: Traffic load/patterns Power Directionality of links Results and analysis to be used as framework for development of new adaptive routing protocol Publications Tan, H. X. Tan and W. K. G. Seah. (2004) Dynamically Adapting Mobile Ad Hoc Routing Protocols to Improve Scalability. Proceedings of the IASTED International Conference on Communication Systems (CSN 2004), Marbella, Spain, Sep 1-3, 2004. Tan, H. X. Tan and W. K. G. Seah. (2004b) Limiting Control Overheads Based on Link Stability for Improved Performance in Mobile Ad Hoc Networks. Submitted to the Second International Conference on Mobile Computing and Ubiquitous Networking (ICMU 2005) *notification of acceptance: 15 Jan 2004. Tan, H. X. Tan and W. K. S. Seah. (2004c) Dynamic Topology Control to Reduce Interference in MANETs. Institute for Infocomm Research Technical Report: NET-2004-AHN-GENERAL-0027, Nov 2004.