IEEE COMMUNICATIONS LETTERS, VOL. 3, NO. 1, JANUARY 1999 1 Logical AND Diversity Combining in FH/SSMA Packet Radio Networks Jik Dong Kim, Student Member, IEEE, and Sang Wu Kim, Member, IEEE Abstract— We propose the logical AND diversity combining in slotted frequency-hopped spread spectrum multiple access (FH/SSMA) packet radio networks. The proposed diversity scheme employs the symbol-by-symbol logical AND operation between the currently received packet and the previously combined packet. Our results show that the proposed lowcomplexity diversity combining scheme provides a significant performance improvement over existing diversity combining schemes. Index Terms—Diversity combining, frequency-hop spread spectrum, multiple access. I. INTRODUCTION P ACKET combining utilizes the additional diversity available in the previous (or failed) packets to achieve the minimum number of transmissions for correct decoding. There are two different approaches in combining packets: symbolby-symbol combining (termed as diversity combining) and codeword-by-codeword combining (termed as code combining) [1], [2]. In this letter we consider the symbol-by-symbol combining. There are a number of papers that study the symbol-bysymbol combining. The combining scheme in [3] assumes that the receiver knows the presence of multiple access interference (MAI), and chooses a symbol that has not been hit by MAI. In [4], soft decision demodulator outputs are added to previously summed outputs at each combining step, and the symbol corresponding to the largest is decided as the transmitted symbol. Li [5] analyzed the performance of ratio-threshold diversity combining (RTDC) in which Viterbi’s ratio-threshold test is performed for each symbol, and the quality bit is utilized in combining. Harvey [6] considered the averaged diversity combining (ADC) which generates the averaged soft decision value based on all received copies, and then considered the weighted diversity combining (WDC) whose weights depend on the channel state. However, these combining schemes require a high computational complexity and large memory. The logical AND diversity combining proposed in this letter is based on the observation that the desired symbol is not changed in all retransmitted packets, whereas the interfering symbols change. Therefore, the desired symbol can be detected from the symbol-by-symbol logical AND operation between the newly received packet and the previously combined packet. Manuscript received May 12, 1998. The associate edtor coordinating the review of this letter and approving it for publication was Dr. B. R. Vojcic. The authors are with the Department of Electrical Engineering, Korea Advanced Institute of Science and Technology, Taejon 305-701, Korea (email: swkim@san.kaist.ac.kr). Publisher Item Identifier S 1089-7798(99)01261-2. The desired symbol produces logic “1,” whereas the interfering symbols produce logic “0” with high probability. II. SYSTEM DESCRIPTION We consider a slotted frequency-hopped spread spectrum multiple access (FH/SSMA) network, where there are transmitter-receiver pairs communicating over frequency Reed–Solomon slots. Each packet is encoded by an , and each -ary code symbol is (RS) code over -ary orthogonal FSK signals. composed of We assume that the frequency hopping is synchronous, and one code symbol is transmitted per hop. The receiver block energy detector diagram is shown in Fig. 1. Each of the outputs is compared with a threshold, , and logic “1” is if the detector output is larger than generated , and logic “0” is generated , the threshold otherwise. The proposed combining scheme is illustrated for the case and in Fig. 2. Let , and of denote the th code symbol in the previously combined packet, the newly received packet, and the newly combined packet, is the logical AND of and . respectively, where and are hit, As illustrated in Ex. 1, even when both logic “1” is generated for the desired symbol and logic “0” for interfering symbols. Thus, the desired symbol can be correctly detected. When more than one logic “1’s” are generated after combining, as illustrated in Ex. 2, we declare an erasure. If we assume that each symbol is transmitted equally likely, , then the probability of correct i.e., with probability detection increases with larger , because the probability that the interfering symbols are identical (producing logic “1”) . Also, as the number of combined decreases with larger packets increases, the probability that logic “1” is generated by the interfering symbols decreases, which leads to an additional performance improvement. III. PERFORMANCE ANALYSIS In this section, we derive the packet error probability in the presence of MAI only. The effect of thermal noise and fading will be examined by computer simulations. In the presence of MAI only, the combined symbol can be either RS code can correct up to erased or correct. Since the erasures [7], the packet containing more than symbol erasures is declared unsuccessful, and a retransmission , where is is requested. If we let the number of packets simultaneously transmitted in the th 1089–7798/99$10.00 1999 IEEE 2 Fig. 1. IEEE COMMUNICATIONS LETTERS, VOL. 3, NO. 1, JANUARY 1999 Receiver block diagram. In (2), is the conditional probability that a modulation symbol is erased after combining, given a hit pattern , and is given by (4) and is the number of logic where (see Fig. 1) for the th “1’s” in denotes the maximum of , i.e., transmitted packet, and . The conditional probability that a tone pattern occurs given a hit pattern , is given by [8] (5) is the conditional probability that a modulation In (4), symbol is erased after combining given a tone pattern , which is given by Fig. 2. Proposed combining algorithm: M = 8; N = 1. transmission, , then the average packet error , after combining packets is given by probability, (1) is the conditional probability that a code where symbol is erased after combining packets, given , which can be represented by (6) represents the number of logic “1’s” in (see Fig. 1) after receiving the th packet , and and are the minimum and maximum of , respectively. is the conditional probability logic “1’s” given and , which is that there are given by [9, p. 60] where (7) (2) and is the number of inwhere is the terfering users in the th transmission, and conditional probability that a hit pattern occurs given , represented by (3) IV. NUMERICAL RESULTS In obtaining numerical results, we consider the case where the number of transmitted packets is fixed in all transmissions ), and that the packet length (i.e., is equal to the code symbol size (i.e., extended RS code). Since ADC in [6] is identical to the scheme in [4] when applied to FH/SSMA systems, we consider WDC in [6]. KIM AND KIM: FH/SSMA PACKET RADIO NETWORKS Fig. 3. Packet error probability PE versus the number of users, no background noise, no fading: q = 50; M = 8, (64, 48) RS code. 3 Fig. 5. Packet error probability PE versus E b =N0 in a frequency-nonselective Rayleigh fading channel with background noise: K = 50, q = 50, L = 3, M = 8, (64, 32) RS code. , used in deciding or 1 (see Fig. 1) is optimized such that the packet error probability is for each minimized. Also, the threshold for Viterbi’s ratio-threshold test in scheme [5] is optimized. We find that the proposed low-complexity combining scheme out performs much more complex alternatives discussed in [3]–[6]. V. CONCLUSIONS Fig. 4. Packet error probability PE versus the number of users, no background noise, no fading: q = 50; L = 3, (64, 48) RS code. We proposed the logical AND diversity combining scheme in slotted FH/SSMA packet radio networks. The proposed lowcomplexity scheme is simple to realize, and gives a (more) significant performance improvement over the existing much more complex schemes as the number of combined packets or the modulation symbol size increases. REFERENCES Fig. 3 is a plot of packet error probability versus the number of users for several values of in the presence of MAI only. Computer simulation is also performed for the proposed logical AND combining scheme to confirm the accuracy of the analytical results derived in Section III. We find that the proposed scheme yields a much lower packet error probability than existing schemes in [3]–[5]. As discussed in Section II, the performance improvement is more significant as increases. versus the Fig. 4 is a plot of packet error probability in the presence of number of users for several values of increases, the packet error probability deMAI only. As creases, as expected. This indicates that the proposed scheme is more effective with larger . versus Fig. 5 is a plot of packet error probability in a frequency-nonselective Rayleigh fading channel with background noise. The packet error probability is evaluated by computer simulations. We assume that fading magnitudes for different users are independent, and the fading magnitude is constant during a modulation symbol period. The threshold, [1] Y. M. Wang and S. Lin, “Modified selective repeat type II hybrid ARQ system: Performance analysis,” IEEE Trans. Commun., vol. COM-31, pp. 593–608, May 1983. [2] D. Chase, “Code combining—A maximum likelihood decoding approach for combining an arbitrary number of noisy packets,” IEEE Trans. Commun., vol. COM-33, pp. 385–393, May 1985. [3] T. Ketseoglou and A. Polydoros, “Code combining techniques for FH/SSRA slotted networks,” in Proc. IEEE MILCOM’90, pp. 344–348. [4] A. M. Y. Bigloo, T. A. Gulliver, and V. K. Bhargava, “A slotted frequency-hopped multiple-access network with packet combining,” IEEE J. Select. Areas Commun., vol. 14, pp. 1859–1865, Dec. 1996. [5] G. Li, Q. Wang, V. K. Bhargava, and L. J. Mason, “Performance of a ratio-threshold diversity combining scheme in FFH/FSK spread spectrum in partial band interference,” in Proc. IEEE ICC’92, pp. 672–676. [6] B. A. Harvey and S. B. Wicker, “Packet combining systems based on the Viterbi decoder,” IEEE Trans. Commun., vol. 42, pp. 1544–1557, Feb./Mar./Apr. 1994. [7] R. E. Blahut, Theory and Practice of Error Control Codes. Reading, MA: Addison-Wesley, 1983. [8] S. W. Kim, Y. H. Lee, and S. M. Kim, “Bandwidth tradeoffs among coding, processing gain and modulation in frequency hopped multiple access communications,” Proc. Inst. Elect. Eng.—Communications, vol. 141, pp. 63–69, Apr. 1994. [9] A. Papoulis, Probability, Random Variables, and Stochastic Processes, Reading, 3rd ed. New York: McGraw-Hill, 1991.