International Journal of Computer Information Systems, Vol. 3, No.

4, 2011

Evaluation of Diversity Combiner in Packet Radio Network System

Kaliyaperumal Karthikeyan* Lecturer, Dept of Computer Science Eritrea Institute of Technology Asmara, State of Eritrea. N.E.Africa E.Mail:kirithicraj@gmail.com M.Aswani Kumar Faculty of ICT Ministry of Education Asmara, State of Eritrea. N.E.Africa E.Mail: aswani. swamy@gmail.com S.K.Manju Bargavi Assistant Professor/Research Scholar Department of Computer Science & Engg., Vidya Vikas College of Engineering Tamilnadu, India E.Mail: cloudbargavi@gmail.com

Abstract - Random access protocols such as the ALOHA are attractive in wireless environments because they do not require any centralized coordination. However, the traditional narrowband ALOHA suffers from its low maximum throughput, inherent system instability, and rapid rising delay under intensive traffic loading. Although its performance can be improved by employing a carrier sensing technique, it is rather difficult to achieve a reliable carrier sensing in a wireless environment due to hidden terminal problems. Besides, there exists a maximum data rate that can be transmitted via a narrowband ALOHA channel (limited by ISI). In order to facilitate high data rate transmission, the only choice until now has been the awkward use of multiple ALOHA channels in parallel. Key Words - ALOHA Channel, Diversity Combining, Throughput, Spread Spectrum, Packet Radio. I. INTRODUCTION Wideband Code Division Multiple Access (WCDMA) is an attractive signaling scheme for high speed packet radio networks because it has good signal capture, along with spread spectrum advantages [1] such as immunity to external interference and jamming, robustness against ISI, anti-multipath capability and low probability of intercept. Consequently, using spread spectrum signaling in random access packet-switched the radio network will provide a possible solution to attain the high throughputs and high data rates required for new applications. However, channel band width utilization levels of CDMA offset, the advantages of packet CDMA.

This is due to the large processing gain required for reliable multiuser operation [2]. Besides multi-user interference, the fading phenomenon of wireless medium severely degrades the system performance. In order to cope with the detrimental effects of channel fading and multiuser interference, some form of diversity reception is needed. Fortunately, a rep eat-request (ARQ) strategy can be easily incorporated into packet switches networks, thereby enabling these systems to operate at higher error rates. Also, the application of Forward Error Correction (FEC) coding can provide significant capacity which increases both in fixed assigned CDMA and slotted packet CDMA. Many authors[2][3] have analyzed the performance of random access networks in CDMA systems. Although a significant improvement in capacity has been observed, the overall (hybrid) system still behaves like familiar ALOHA in the sense that the system fails when loaded beyond a certain point. Slotted ALOHA random access was then used with CDMA for channel access and a type-I hybrid ARQ protocol was employed for error control[4] [5] . Though such system has been shown to have a good network utilization and reasonably high efficiency, relatively high loads and high jammer levels, it suffers from the same disadvantages as type-I hybrid ARQ protocol. In the conventional random-access packet switched DS/CDMA (hereafter, simply referred to as DS/CDMA ALOHA) [2] , erroneous packets will be discarded, and decoding of a newly received copy is attempted at the receiver without using any prior knowledge from previously received copies of the same pack et (which were not decoded successfully). Discarding the noisy packets seems to be wasteful, particularly when one realizes that a portion of a packet may have been received correctly. Chase [6] suggested combining the noise corrupted packets at the codeword level to form a single more reliable packet, thus minimizing the average number of re-transmission required before a packet is

October Issue

Page 97 of 116

ISSN 2229 5208

received successfully. This scheme is adaptive only in minimum number of packets necessary to form a reliable packet transmission. The average number of retransmission is minimized, that allows a greater throughput while maintaining the reliability of the accepted data. Packet combining can therefore be readily applied to a CDMA packet radio network to improve the system capacity and reliability. In [7], packet combining technique applied to a slotted frequency hopped spread spectrum multiple access (FH/SSMA) system. Significantly, performance improvement was shown over the no combining and the narrowband ALOHA cases. It should be pointed out that in [7] the authors have neglected the effects of channel fading and additive white Gaussian noise in their analysis, by assuming co-channel interference as the only source of error. Furthermore, only in lower bound the throughput performance (worst-case analysis) of the new system was presented. Reference [8] examined the use of average diversity combining for asynchronous DS/CDMA packet radio network, operating over AWGN channel. In [9], the authors improved their previous work to present the results of average diversity combining scheme for mobile fading channels. However, it must be pointed out that in [9], the authors obtained the system throughput results of the proposed system through computer simulations and no analytical expressions of the upper and lower bounds on throughput were presented. In contrast, the performance of EGC (referred to as average diversity combiner in [8] and [9]) is evaluated for slotted random access asynchronous DS/CDMA radio networks with retransmission diversity combining. Upper and lower bounds on the system throughput and the probability of bit error are derived for Rayleigh, Nakagami-m and Rician multipath fading environments. In our system, the combining operation occurs at bit level, and the resulting bit decisions are produced by employing equal gain diversity at the output of the rake receiver. Specifically, a novel mechanism is to ensure a reliable packet combining operation is outlined here and the effects of the timevarying multiuser interference (due to the bursty nature of the traffic) on the system performance is investigated by validating the tightness of the upper and lower bounds for the throughput and packet error probability. Finally, the sensitivity of the system performance to the channel impairments, such as the packet header failure and the missed acknowledgement messages is also studied.

International Journal of Computer Information Systems, Vol. 3, No. 4, 2011 II. DS-CDMA PACKET RADIO NETWORK ANALYTICAL MODEL When the spread-spectrum multiple access channel is operated in a random access packet mode analysis which proceeds analogous to the analysis of ALOHA channels, with the difference that m >1 simultaneous transmissions do not necessarily produce a collision, but results in a soft degradation in error rate. The assumptions made here are similar to those adopted in [3], [4], [6] and [ 7]. A schematic representation of traffic flow in a random access DS/CDMA system with packet combining is shown in the Fig1. Each transceiver can be one of the two operation modes either origination mode or backlog mode. In the origination mode, there is no backlogged packets to be retransmitted and new packets are transmitted in any given slot with probability Po. When a packet retransmission is requested, the user enters the backlogged mode, where the probability of re-transmission of this packet in the next slot is geometrically distributed with probability Pr. If P0 = Pr = P the composite packet arrival distribution becomes binomial with parameters P and Kmax (Number of subscribers). Further simplification occurs when P0 = Pr = P 0 and Kmax because the binomial distribution approaches the Poisson traffic model with an arrival rate equal to the traffic intensity.

Fig 1: Schematic representation of traffic flow in the proposed random access DS/CDMA system. The composite packet arrival distribution is [2],

. (1)

October Issue

Page 98 of 116

ISSN 2229 5208

with a finite composite arrival rate and packet reception interval T. Hence = T offers traffic (average number of attempted transmissions per slot). Modeling the time varying nature of the multiple access interference is simplified if the bit error probabilities in adjacent bits of a packet are independent of each other. This condition is satisfied when the spreading signature sequences are truly random and if perfect chip level synchronization is achieved. The second requirement is a constraint that is needed for analytical purposes and tend to impact the achievable performance adversely. However, the statistical independence assumption become quite realistic if the system is operating in fast fading environment or in an ideally interleaved condition. Although the latter delays or latency ramifications. These are less critical than those due to retransmissions. Furthermore, packet combining can reduce the correlation and the occurrence of bit error random. It is also reasonable to assume that errors in different packets occur independently, since the CDMA spreading codes generally pseudorandom to make this allowance. When the information bits from a particular user are ready for transmission, bits for error detection based on a high error detecting code C0 are appended in yielding an information sequence of Kd bits. Subsequently they are encoded for error correction by a different code C1(e.g.,(Nd, Kd, Td)BCH code) to form a packet of Nd. Let us assume that the same cyclic code will be used for both error correction and error detection. The FEC code is capable of correcting Td random errors per packet. The encoded binary data is then phase modulated to yield a constant envelope datamodulated carrier before a direct-sequence spreading operation. Similar operations are performed by K other active users in the system. The transmitted signal is perturbed by selfinterference and channel attenuation due to multipath fading, multiuser interference, and additive white Gaussian noise. In order to capture the signal from user i , the received composite signal is assumed to go through a rake receiver with M taps, where M = L and L correspond to the maximum number of resolvable multipart. First, the received signal is dispread independently for each multipath component by multiplying the spreading signature code of the i th user, which is delayed by an amount equal to the estimated multipath delay. The modulated signal is then down converted to a base band signal and passed through a bank of correlates prior to diversity combining. At the receiver, EGC technique is used here to combine both the packets and the multipath signals. In equal gain combiner at the receiver, the voltage signals

International Journal of Computer Information Systems, Vol. 3, No. 4, 2011 from each of the M path diversity branches are first cophased, and then summed together. It is assumed that the feedback channel error can only make ACK and NAK messages indistinguishable and the transmitter will handle this error ACK or NAK message(erasure symbol) as a NAK. Such an assumption is reasonable since the undetectable error probability can be made very small by proper selection of error detecting codes. The following two additional protocols are required for a noisy feedback channel: (a) Each time the transmitter sends out a packet, a timer for the packet is initialized and started. If the response from the receiver is not obtained for the packet after a reasonable period of time, it is assumed that the response is a NAK; (b) If the receiver receives a packet that has already been accepted, an ACK is sent to the transmitter and the packet is discarded. III. PACKET IDENTIFICATION MECHANISMS If the pseudo-random (PN) spreading code is unique for each user, the signal will appear as random noise for unintended recipients. This protection level (also known as processing gain, PG) is described by the ratio between the PN sequence chipping rate and the encoded source rate. It is evident that the spreadingdespreading operation helps to reduce or eliminate the probability of wrongful combining of packets from different users in a receiver-oriented protocol. Let us consider the packets receiver by a particular user. The receiver is expected to determine if a received packet is a new packet or a retransmission of a previously transmitted packet before it is routed to the combiner. In a practical operating environment, the receiver may not be able to determine the packet sequence number. It is now completely free of errors or contain a correctable error pattern. As a consequence, the receiver may incorrectly combine two or more different packets intended for the same user. In this case, there is obviously no reason for additional diversity. Hence, it is evident that the results presented in [7] and [8] are contingent upon the perfect knowledge about the sequence numbers of the noisy packets. IV. EXTRA PROTECTIONS FOR THE PACKET HEADER Since the packet header contains at least a source identification and packet sequence number, which are used in packet combining process. It is natural to protect this header information (Kh bits) by using a powerful FEC. In our scheme, the message itself is contained within a packet, with source address and packet sequence number preceding the message in a

October Issue

Page 99 of 116

ISSN 2229 5208

header , as shown below in Fig. 2. The header is encoded separately from the message so that header may be read before making a decision as to whether to read the following message action. When the packet header preceding a message is not decoded correctly, then the packet (i.e., message) will be discarded and a retransmission will be requested. However, if the packet header is detected correctly but not the message, then the packet combining operation will be triggered. An ACK will be sent on to the return channel if both packet header and the message were received successfully. It is worth noting that the likelihood of a packet header being missed, the corresponding message block being received correctly is negligible because the header is normally protected with a low rate code and its block size is usually much smaller than the message block length, i.e., Nh >> Nd.

International Journal of Computer Information Systems, Vol. 3, No. 4, 2011 the offered load Kmax corresponds to the maximum number of users that can be simultaneously handled by the system because the number of receivers or available code words is limited. Exact evaluation of the throughput performance with packet combining is rather involved, specifically because the packet success probability is dependent not only on the most recently received packet, but also related to previously received copies of that packet. In particular, the time-varying co-channel interference produces a nonstationary packet error probability. This is true for both slotted and unslotted radio networks under consideration. Since the time-varying nature of multiuser interference imposes considerable analytical and computational difficulties, we resort to bounding techniques to investigate the system performances. Subsequently, the tightness of the upper and lower bounds for the frame-error rate and throughput are validated using the approach given in [10]. VI. BOUND ESTIMATION FOR PACKET ERROR PROBABILITY Let (K1, K2.Ki) be the event that after combining i copies of the same pocket. decoded data block still contains an uncorrectable error pattern, conditioned on Kj + 1 simultaneous users during the transmission of the j the copy j= 1,2,..i. The probability of this event is denoted as Pi = p{D(i) (K1, K2,Ki) } Since a decoding failure of the combined packet after the i the reception will initiate the request of an i+1th transmission of that packet, it is evident that the packet error rate with i number of transmissions PB(i) is equal to the repeat request probability (RRP) of the i+1th transmission: . (3) For instance, PB(1) corresponds to the RRP of the first retransmission (i.e., the probability that original packet contains an uncorrectable error pattern), while PB(2) denotes the RRP of the second retransmission.

Fig 2: Data frame format for: (a) Conventional DS/CDMA ALOHA (i.e., without packet combining); and (b) Packet combining systems. V. SLOTTED DS/CDMA ALOHA -THROUGHPUT ANALYSIS: The average packet error probability or average bit error rate metric is related to a particular link performance. whereas throughput and average time delay is representative of network performance in terms of channel bandwidth utilization efficiency and stability. As in [2], the steady-state throughput is defined as the expected number of successful transmissions per slot. Then, the throughput of a conventional slotted random access DS/CDMA is given by, . (2) Where Pc(m) denotes the packet success probability conditioned on m=K+1 simultaneous transmissions during the packet transmission interval, is

Evaluation of Eq. (3) is rather difficult because the events p{D(1) (K1)}, p{(D(2) (K1,K2)}, ,p{D(i) (K1,K2,Ki)} are not independent. However, it is possible to derive the upper and lower bounds. The RRP of the i +1 the transmission can be rewritten as

lower

.. (4) where the upper bound is given by Pi, the bound can be obtained by assuming that

October Issue

Page 100 of 116

ISSN 2229 5208

P1,P2,Pi are all independent events. Using the upper and lower bounds for the RRP, the packet error probability is bounded by,

International Journal of Computer Information Systems, Vol. 3, No. 4, 2011 direct-sequence CDMA networks, IEEE Trans. on Selec. Areas in Communications, SAC-8 (4), May 1990. [5] S. W. Kim, Optimum rate Reed-Solomon codes for frequency- hopped spread-spectrum multiple access systems, IEEE Trans. On Communications, COM-37, No. 6, pp. 138-144, Feb. 1989. [6] D. Chase, Code combining- A maximum likelihood decoding approach for combining an arbitrary y number of noisy packets, IEEE Transactions on Communications, Vol. 33, pp. 385-393, May 1985. [7] A. M. Y. Bigloo, T. A. Gulliver and V.K Bhargava, The slotted frequency hopped multiple access network with packet combining, IEEE Journal on Selected Areas in Communications, Vol. 14, pp. 1859-1965, Dec 1996. [8] S. Souissi and S. B. Wicker, Performance analysis of a random access direct sequence, packet combining, code division multiple access system, IEEE GLOBECOM, pp. 144-148, Dec. 1994. [9] S. Souissi and S. B. Wicker, Performance analysis of a random access direct sequence packet combining code division multiple access system, IEEE Transactions on Vehicular Technology, Vol. 44, pp. 304 -312, May 1995.
AUTHORS PROFILE Mr.Kaliyaperumal Karthikeyan is working as a Lecturer in the Department of Computer Science, EIT, Asmara, State of Eritrea, affiliated to the Ministry of Education. He has blended his wide experience of 15 years in teaching, research in the field of CS and Electronics. He has published a number of international journal papers & books related to the Compute Science. IAENG.

. (5) Where PE (Kx) = p{D(1) (Kx)} note that the the block error rate of the conventional system (type-I hybrid ARQ) is given by Pb(i) = PE(K1) x PE(K2) xx PE(Ki), which is equivalent to the performance of simple time-diversity strategy without packet combining. VII. CONCLUSION A steady-state performance analysis has been presented for slotted random access packet switched DS/CDMA with packet combining, using a two-dimensional diversity equal gain receiver. The packet combining technique is highly advantageous for systems, which can tolerate a certain delay and operate over highly timevarying channels. It also dramatically reduces the average number of transmissions required before a packet is received correctly, thereby improving the delay performance of type-I hybrid ARQ system. Simple and computationally efficient formulas are used for evaluating the average number of transmissions and throughput of this new system with Poisson traffic were derived. Numerical results reveal that the performance of proposed adaptive retransmission diversity combining system with EGC, while being comparable to that of MRC, yields a higher maximum throughput than a conventional DS/CDMA ALOHA . It also sustains the high level of throughput over a wide range of traffic. This performance improvement was obtained at the expense of only very little increase in implementation complexity. REFERENCES [1] M. B. Pursley, The role of spread-spectrum in packet radio network, Proc. IEEE, Vol. 75, pp. 116-134, Jan 1987. [2] D. Raychaudhuri, Performance analysis of random access packet-switches code division multiple access, IEEE Trans. On Communications, COM-29, No. 6, pp. 895-901, June 1981. [3] A. Polydoros and J. Silvester, Slotted random access spread-spectrum networks: An analytical framework, IEEE Journal on Selec. Areas in Communications, SAC5, pp. 989-1002, July 1987. [4] J. M. Hanratty and G. L. Stuber, Asymptotic performance analysis of hybrid ARQ protocols in slotted

. He is also the member of

Mr.Aswani Kumar is working as a Faculty of ICT, Ministry of Education., Asmara, Eritrea. He has blended knowledge of research work in the field of Computer Science and Physics He has published a number of international journal papers related to the Compute Science.

Mrs. Manju Bargavi is working as a Asst. Professor in the Department of Computer Science & Engineering. Vidya Vikas College of Engineering, Tamil nadu, India. She is research scholar in Anna University, Coimbatore. She has blended the wide experience of Ad hoc and networking research in the field of Computer Science & Engineering.

October Issue

Page 101 of 116

ISSN 2229 5208