Adaptive Sub-block ARQ techniques for wireless networks by ijcsis


More Info
									                                               (IJCSIS) International Journal of Computer Science and Information Security,
                                               Vol. 8, No. 6, September 2010

         Adaptive Sub-block ARQ techniques for wireless networks.
                 A.N.Kemkar, Member, ISTE and Dr. T.R.Sontakke Member,ISTE

                                                              delays but is sensitive to loss and errors. Many
Abstract— Wireless channels are highly affected
                                                              researchers have devoted much effort to analyze
by unpredictable factors such as co channel
                                                              the throughput for various ARQ protocols in
interference, adjacent channel interference,
                                                              Rayleigh fading channels. In the packet data
propagation path loss, and shadowing and multi
                                                              transmission, short packets are less likely to
path fading. An adaptive ARQ scheme, we mean
                                                              encounter fades than long packets, but on the
an ARQ scheme with two or more different
                                                              other hand, they are more burdened by overheads.
transmission modes meant for different channel
                                                              In other words, the packet length to maximize the
conditions, which uses some channel sensing
                                                              instantaneous throughput is closely related to the
mechanism to decide which transmission mode is
                                                              dynamic channel condition due to fading,
used. In this paper, we propose an adaptive
                                                              shadowing, and propagation path loss. Therefore,
transmission of sub-blocks scheme, for wireless
                                                              if we choose the optimum packet length
networks. As the channel becomes increasingly
                                                              adaptively by estimating the channel condition,
noisy, the data block is divided into smaller sub-
                                                              we can continuously achieve the maximum
blocks for transmission. Each sub-block is
                                                              attainable throughput.
encoded for error control by a CRC code. The
                                                              Paper is organized as follows. We start by,
received block is checked for errors sub-block by
                                                              describes in detail about related work in section 2.
sub-block. The propose sub-block retransmission
                                                              A Communication system model and proposed
scheme provides improved throughput over
                                                              method of adaptation in section 3. In section 4
conventional ARQ schemes by retransmitting
                                                              we have presented system analysis 5. Simulation
only the same number of sub-blocks in the
                                                              parameters and results . Followed by conclusion
occurrence of errors. .
                                                              in section 6.
.Index Terms—Retransmission protocol; Hybrid
ARQ,CRC                                                       2. RELATED WORK:
1. INTRODUCTION:                                              A change of transmission mode can mean, for
In a mobile radio channel, burst errors frequently            example, a change of the packet size in the SR
occur because of Rayleigh fading and shadowing.               scheme [1] or a change of the number of
In particular, for a large cell-size system with the          transmitted copies of a packet in the GBN scheme
radius of more than several km, shadowing often               [2] or a change of the code rate in an HARQ-I
becomes the predominant source of burst errors.               scheme [3]. In these schemes, the channel sensing
(Shown in Fig-1)                                              is    usually      done      by     observing     the
Therefore, in order to provide reliable packet data           acknowledgements sent by the receiver to the
transmission in such a channel, an efficient                  transmitter. This can mean either estimation of
automatic repeat request (ARQ) protocol must be               error rates, as in [4], or detection of channel state
employed, since data service can tolerate some                changes, as in [5] and [6],which does not require
                                                              as long an observation interval (OBI) as reliable
A.N.Kemkar1 ,S.R.T.M.U,Nanded.                               error rate estimation.
+91-9819150392,                             In [2], an adaptive SR scheme was proposed,
Dr.T.R.Sontakke2                                              where the packet size used in the current
Ex.Director – S.G.G.S.I.T.E.- Nanded                          transmission was selected from a finite set of
Principal,Sidhant college of Engineering                      values based on a long-term BER estimate. This
Pune.+91-9822392766,                      estimate was obtained by counting the incorrectly
                                                              received packets over a time interval and

                                                                                          ISSN 1947-5500
                                                (IJCSIS) International Journal of Computer Science and Information Security,
                                                Vol. 8, No. 6, September 2010

assuming that there can be at most one bit error in            non coherent detection (non coherent FSK)
an erroneous packet. Another adaptive SR                       scheme in terms of easy implementation, because
scheme with variable packet size was proposed in               it requires no complicated carrier recovery circuit.
[5], where the a posteriori distribution of the
BER was computed based on the number of
retransmissions during the OBI, and the packet
size was selected so that the expected efficiency                                                       Transmitter
of the protocol was maximized. In [2], Yao                                                              Electronics
proposed an adaptive GBN scheme where the
transmitted number the transmitted number of
copies of a packet was variable.
Numerous adaptive HARQ schemes have been                                           Adaptive Logic
suggested in the literature. Typically, the code
rate is varied according to the estimated channel
conditions. In [6] and [7], adaptive HARQ-I                                                             Receiver
schemes were studied with convolutional codes                               Buffer
used for error correction. Finite-state Markov
models were assumed for the channel. Switching
between transmission modes depended on the                     Fig.1. Communication System model.
number of erroneous blocks occurring during an
OBI. A similar adaptive HARQ-I scheme with                     Table 1 –Communication system parameters
either block or convolutional codes were
proposed in [3]. In [4], sequential statistical tests          Parameters                      Description
                                                               Channel type                    Contention Free, half duplex
were applied on the acknowledgements to detect                 Modulation                      Binary FSK,Non coherent
channel state changes. An adaptive HARQ-II                     /Demodulation                   detection.
scheme with variable packet size was proposed                  Packet structure                Information packet length
for wireless ATM networks in [8]. This scheme                                                  256,128,64,32,16 bits
used rate compatible convolutional (RCC) codes
for error correction. In [9], three different                  .
adaptive HARQ schemes are proposed using                       3.2. Adaptation Policy
Reed-Solomon codes for error correction.                         According to the variations of SNR, the receiver
Another adaptive HARQ scheme using Reed-                       channel may be consider to be in one of the states
Solomon codes with variable rate for error                     at each instant ‘ t ’. We assume that the sender
correction was proposed in [10]. In this scheme                knows the state at the transmission time for
short-term symbol error rate was estimated by                  receiver. Let’s define the transmission status at
computing the bitwise modulo-2 sum of two                      time t as the set of all channel state. Before
erroneous copies of a packet. This method was                  transmitting, the adaptive algorithm in the sender
originally proposed in [11].                                   must estimate the efficiency and packet loss rate
3. A COMMUNICATION SYSTEM MODEL                                of the ARQ/FEC protocol using all the available
AND PROPOSED METHOD OF                                         coding schemes as well as the ARQ protocol as a
ADAPTATION:                                                    function of the transmission status. It then tries to
3.1 A communication system model:                              find the protocol satisfying the desired packet loss
Fig. 1 shows the communication system model.                   rate. The time is divided into transmission rounds.
In the non cellular or large cell-size system, a               Each transmission round corresponds to the
radio base station continuously transmits data                 transmission ‘k’ packets in case of ARQ. A
packets to a single mobile terminal with no                    transmission round ends when the sender is
packet collision after the link connection is                  informed about the reception state of the receiver.
established. Table I summarizes the digital                    The adaptive algorithm is repeated at the end of
mobile communication system characteristics,
where we choose a binary frequency shift keying

                                                                                           ISSN 1947-5500
                                               (IJCSIS) International Journal of Computer Science and Information Security,
                                               Vol. 8, No. 6, September 2010

each transmission round. Basically, the algorithm             as a retransmission units with fixed rate of data
goes through the following steps:                             transmission. Hence we can send the packet of
1. At the beginning of the algorithm, the sender              specific size based on the estimated signal to
determines the desired packet loss rate (SNR) of              noise ratio (SNR).As shown in the Fig.1 the
the session. It also determines the transmission              system configuration of ARQ techniques
status.                                                       combines with adaptive packet size modulation.
2. The sender estimates the packet loss rate of the           With an exact bit error rate equation for FSK at
ARQ protocol as well as the ARQ/FEC protocol                  certain signal to noise ratio ‘ γ ’, we can decide
using all the available coding schemes, based on              the value of packet size satisfying the required
the transmission status. It then adjusts its                  BER (bit error rate) Assuming that we have ‘ M
parameters and starts the transmission of the                 ’different block sizes { L1L 2 L 3 L 4......L M } .
3. At the end of a transmission round, the sender             Let Ai; i ∈ { 0,......M } with Ai as the threshold
again determines the transmission status. It then             value of signal to noise ratio, being between the
                                                              i level and i + 1 level.
repeats the step.
                                                              A0 is the lowest possible signal to noise ratio and
4. PERFORMANCE ANALYSIS OF THE                                A M is the highest possible signal to noise ratio.
The performance analysis of the scheme is                     5. SIMULATION RESULTS:
measured in terms of throughput of the proposed                 We evaluate the performance of the proposed
scheme. Further we show the comparison of                     scheme implemented with Matlab. We run the
throughput with sub block and without sub block               simulation for three schemes i.e. with sub block
transmission schemes with Adaptive scheme.                    transmission and without sub block transmission
Expression of throughput for ARQ for present                  and adaptive. The simulation parameters are
scheme:                                                       shown in the table 2.
η=             (1)
     E[T]                                                      Table 2- Simulation Parameters
Where K =information bits in a block. E [ .]
                                                              Parameters                         Notation Values
=Expectation of number of transmitted bits in a               Signal to Noise Ratio                  γ    Varied
given block.                                                  Threshold values of                snr      8,7,6,5
T = Mn + ∑ T i           (2)                                  SNR
           i =1                                               Max. number of                     --------        3
 Where M =number of sub blocks, n =number of                  Retransmissions
bits in a sub block, T i =number of transmitted bits          Number of sub blocks                L              Varied
for i th transmission.                                        retransmitted
                  ∞                                           Cyclic Redundancy                  CRC             Varied
E [ T ] = Mn + ∑ E [ T i ] (3)                                Check
                  i =1                                        Bit error rate                     BER             Varied
Where E [ T ] =Average number of transmitted                  Packet error rate                  PER             Varied
                                                              Throughput efficiency              η               Varied
Out of M sub blocks if L sub blocks are                       Data Rate                          R               9.6kbps
transmitted at the i th retransmission, then random
                                                                   Simulation runs for 5000 total blocks. Result
variable, T i takes the value Ln ,if L out of M
                                                              is the average of independent experiments where
sub-blocks are retransmitted at the i th                      each experiment uses different randomly
retransmission. Our algorithm compute the value               generated uniform parameters. We use mean
of (3) to get the result from equation (1).In our             values which are obtained independent
analysis we have consider the variable packet size

                                                                                          ISSN 1947-5500
                                                                      (IJCSIS) International Journal of Computer Science and Information Security,
                                                                      Vol. 8, No. 6, September 2010

experiments as a basic data to get the result.
Simulation results are shown Table 3.

    For the packet error rate 0.3,0.5,0.7 the
    throughput of the system is say η 1,η 2,η 3 .
                                                                                                                                   T h ro u g h p u t V s . P a c k e t E rro r R a te

    Adaptive Scheme                                 Throughput efficiency                                   0 .9
    SNR   Sub                            Packet     η1       η2       η2                                    0 .8
          block                          size                                                               0 .7
    8     2                              256        0.92     0.84     0.82
    7     4                              128        0.93     0.85     0.83                                  0 .6
    6     8                              64         0.94     0.92     0.88

                                                                                        T h ro u g hp u t
                                                                                                            0 .5
    5     16                             32         0.97     0.95     0.93
    Below 32                             16         0.98     0.94     0.96                                  0 .4
                                                                                                            0 .3

    Table 3.Simulation Results of adaptive                                                                  0 .2
                                                                                                                                                         W it h M ic r o b lo c k
    scheme                                                                                                                                               W it h o u t M ic r o b lo c k
                                                                                                            0 .1
                                                                                                                                                         A d a p t iv e A R Q
                                                                                                                 0    0 .1 0 .2 0 .3 0 .4 0 .5 0 .6 0 .7 0 .8 0 .9            1
                                                                                                                                     P a c k e t E rro r R a te
                                         T h r o u g h p u t V s . P       a c k e t E r r o r R a t e
                                                                                     Fig. 3. Comparison of three scheme
                                                                                     6.CONCLUSION:                            Table -2 and Fig.2 shows the
                          0 . 9 5
                                                                                     result                          of     five     modes        of     sub      block
                                                                                     transmission .Proposed Adaptive scheme
                          0 . 9                                                      will choose proper mode in accordance with
                                                                                     the channel parameters, here                                            threshold
     T h ro u g h p u t

                                                                                     values of SNR (considered as channel
                          0 . 8 5
                                                                                     sensing mechanism).From the Fig.3 the
                                       2 S u b B lo c      k                         simulation results showed that the simulated
                                       4 S u b B lo c      k
                          0 . 8                                                      throughput points                                    for adaptive scheme
                                       8 S u b B lo c      k
                                       1 6 S u b B lo      c k                       follows                              the ideal throughput curve(with
                                       3 2 S u b B lo      c k
                                                                                     micro block transmission) very closely. The
                          0 . 7 5
                             0 0 . 10 . 20 . 30 . 40 . 50 . 60 . 7 0 . 80 . 9 1      proposed Adaptive sub-block retransmission
                                              P a c k e t E r r o r R a t e
                                                                                     scheme improved the throughput and the
                                                                                     reliability by using dynamically adapting the
    Fig. 2. Comparison as per Table -2                                               number of Sub block transmission according

                                                                                                                                       ISSN 1947-5500
                                         (IJCSIS) International Journal of Computer Science and Information Security,
                                         Vol. 8, No. 6, September 2010

to length to the varying channel packet error                    adaptive error control. IEEE
                                                                 Transactions       on      Vehicular
                                                                 Technology,            50:1426–1436,
REFERENCES :                                                     November 2001. IEE Proceedings,
                                                                 Part F, 135:85–94, February 1988
    [1] J.A.C. Martins and J.C. Alves. ARQ
                                                             [11]           S.S. Chakraborty, M.
        protocols with adaptive block size
                                                                 Liinaharja, and E. Yli-Juuti. An
        perform better over a wide range of
                                                                 adaptive ARQ scheme with packet
        bit error rates. IEEE Transactions on
                                                                 combining      for    time   varying
        Communications, 38:737–739,June
                                                                 channels. IEEE Communications
                                                                 Letters, 3:52–54, February 1999.
    [2] Y.-D. Yao. An effective go-back-N
        ARQ scheme for variable-error-rate
        channels. IEEE Transactions on
        Communications, 43:20–23, January
    [3] M. Rice and S.B. Wicker. Adaptive
        error control for slowly varying
        channels. IEEE Transactions on
        Communications,          42:917–926,
        February/March/April 1994.
    [4] M. Rice and S.B. Wicker. A
        sequential scheme for adaptive error
        control     over    slowly    varying
        channels. IEEE Transactions on
        Communications,        42:1533–1543,
        February/ March/April 1994.
    [5] E. Modiano. An adaptive algorithm
        for optimizing the packet size used in
        wireless ARQ protocols. Wireless
        Networks, 5:279–286, July 1999
    [6] B. Vucetic. An adaptive coding
        scheme for time-varying channels.
        IEEE           Transactions        on
        Communications, 39:653–663, May
    [7] B. Vucetic, D. Drajic, and D. Perisic.
        Algorithm for adaptive error control
    [8] I. Joe. A novel adaptive hybrid ARQ
        scheme for wireless ATM networks.
        Wireless Networks, 6:211–219, July
    [9] S. Choi and K.G. Shin. A class of
        adaptive hybrid ARQ schemes for
        wireless links. IEEE Transactions on
        Vehicular Technology, 50:777–790,
        May 2001.
    [10]        H.Minn,M. Zeng, and V.K.
        Bhargava. On ARQ scheme with

                                                                                    ISSN 1947-5500

To top