APTITUD DEL BUQUE PARA CONSERVAR LA FLOTABILIDAD *13 Y UBICACION DE LOS TANQUES DE CARGA

An uplink synchronous MC-CDMA system with K users with equal number of subcarriers and spreading length of N = 32 is considered. Note however that the number of subcarriers need not be equal to N and the equality is assumed here for simplicity in performance evaluation only. The system is simulated in MATLAB using 10000 BPSK data symbols for each user with i.i.d. Rayleigh flat fading on each subcarrier with normalised Doppler rates of fdTb = 0.003 and 0.01,

respectively. For the evaluation of MRC receivers, perfect channel estimation is assumed (although unrealistic this is assumed here for comparison purpose only). Two types of spreading sequences with N = 32 are used: orthogonal Walsh sequences and Gold sequences [89]. The Gold sequences are obtained from two fifth order polynomials with octal values 45 and 75, giving the sequence length of 31 and single random bits are appended to make the sequence length N = 32 for the fair performance comparison with Walsh sequences.

BER vs.Eb/N0

The BER performance of the BASC technique is compared with EGC and MRC under the same system loading of K = 10 users is shown in Figure 3.31 for Walsh sequences. BASC with µ = 0.003 is selected for reasonable adaptivity. As expected, it shows significantly improved BER compared with EGC and MRC. MRC shows degraded BER than EGC under Walsh sequences and approaches EGC in the case of Gold sequences. The MRC receiver shows BER degradation com- pared with EGC [71] and the proposed technique. It can be noted that the BASC receiver restores some of the orthogonality of Walsh sequences used and hence achieves improved BER. Similarly BER performance using Gold sequences are shown in Figure 3.32. BASC continue to show im- proved BER compared with EGC and MRC while the performance of EGC approaches that of MRC. All schemes show less performance with Gold sequences due to higher cross-correlation. In higher mobility scenario with fdTb = 0.01, BASC shows slightly less performance gain, due

to insufficient channel tracking. In Figure 3.33, the BER of three techniques are compared under the same system load of K = 16 using Gold sequences and compared with Walsh sequences and fdTb = 0.003. The proposed receiver shows BER much improved compared with EGC and MRC.

For example at the Eb/N0 = 15dB, the BER of 0.0065 is achieved compared with 0.023 and

0.031 for the other receivers. The use of Gold sequences shows slight degradation compared with Walsh sequences at higher SNR region i.e. Eb/N0 > 15dB. This result provides some insights

into operation of the BASC under asynchronous environments where non-orthogonal sequences such as Gold or PN sequences are normally used.

User Capacity

Figure 3.31: BER vs. Eb/N0performance of Blind Adaptive Subcarrier Combining for uplink MC-CDMA

with K=10, µ = 0.003, and Walsh sequences of N=32

Figure 3.32: BER vs. Eb/N0performance of Blind Adaptive Subcarrier Combining for uplink MC-CDMA

Figure 3.33: BER vs. Eb/N0performance of Blind Adaptive Subcarrier Combining technique for uplink

MC-CDMA for K=16 , µ = 0.003; Walsh (squares) and Gold (triangles) sequences of N=32

ure 3.34. It can be clearly seen from the figure that the BASC receiver shows much higher user capacity compared with other techniques. For example, at the same BER of 0.01 the proposed technique can support 19 users compared with 15 of EGC and less than 8 users with MRC. Simi- larly the techniques are compared with the use of Gold sequences under the same settings in Figure 3.35. The BASC receiver shows significantly higher user capacity under this system settings. For example, at the same BER of 0.01 the BASC receiver can support 17 users compared with 4 of EGC and 7 users with MRC, confirming the significant gain of the proposed blind adaptive design approach. Interestingly MRC approach shows improved performance under low system loading compared with EGC and gradually degrades as loading increases (K > 8). Again, in higher mo- bility scenario with fdTb = 0.01, BASC supports slightly less users, due to insufficient channel

tracking.

The effects of step-size

More interesting aspects of the proposed technique is now investigated. The effect of step-size on the performance of the proposed blind adaptive receiver under system loading of K = 10 and 16 under the same Eb/N0= 20 dB are presented. In Figure 3.36 the BER results of the proposed

blind adaptive receiver is investigated under the range of step-sizes µ = 0.0003−0.04 using Walsh sequences. It can be clearly seen from the figure that the proposed technique provides significant BER improvement under all step-size used. The vertical line with stars presents BER under the step size µ = 0.003 used in earlier simulations. Interestingly, the optimum step-size for K = 10 and 16 are shown to be µ = 0.025 and 0.02, respectively. With the use of optimum step-size for

Figure 3.34: BER vs. number of users performance of Blind Adaptive Subcarrier Combining for uplink MC-CDMA, Eb/N0= 15dB, µ = 0.003, and Walsh sequences of N=32

Figure 3.35: BER vs. number of users performance of Blind Adaptive Subcarrier Combining technique for uplink MC-CDMA, Eb/N0= 15dB, µ = 0.003, and Gold sequences of N=32

K = 10, significant improvement in BER is shown to be achievable with the proposed receiver compared with EGC and MRC. Similar BER performance improvement is also seen for the case of K = 16.

Figure 3.36: Effect of step-size µ on the BER performance of Blind Adaptive Subcarrier Combining for uplink MC-CDMA with K=10 and 16, Eb/N0= 20dB, and Walsh sequences of N=32

In Figure 3.37 the BER results of the proposed blind adaptive receiver with Gold sequences is investigated under the range of step-sizes µ = 0.0003 − 0.04. The proposed technique provides significant BER improvement with most step-sizes used. The vertical line with stars presents BER under the step size µ = 0.003 used in earlier simulations. The optimum step-size for K = 10 and 16 are shown to be µ = 0.015 and 0.01, respectively. For example, with the use of optimum step-size for K = 10 it achieves significant BER improvement compared with EGC and MRC giving 0.0004. Similar gains are also shown for K = 16.

Figure 3.37: Effect of step-size µ on the BER performance of Blind Adaptive Subcarrier Combining for uplink MC-CDMA with K=10, Eb/N0= 20dB, and Gold sequences of N=32

3.5

Chapter Summary

This Chapter is focused on maximizing the number of users in uplink CDMA by designing more advanced receiver techniques and algorithms. In Section 3.2, an improved CMA-SIC receiver is proposed for DS-CDMA, using CM algorithm embedded in each SIC stage to perform the user’s amplitude estimation for the detection and cancellation. It showed good performance improve- ment in fading and near far conditions by combining MAI suppression capabilities of the CMA and SIC. At reasonable system loading of 10 users, the proposed design structure gave the same performance of a single user. Much improved user capacity compared with conventional SIC and CMA and MF receivers are shown to be achieved. The effect of step-size on the BER is also investigated.

A novel blind multistage PIC receiver referred to as BA-PIC for uplink DS-CDMA is pro- posed in 3.3, exploiting the constant modulus property of users’ signals. An effective algorithm for blind interference suppression for detection and MAI amplitude estimation for the cancella- tion is also proposed to mitigate the error propagation and fading channel estimation problems of conventional PIC. Besides being blind and low complexity PIC, it showed significantly improved performance compared to conventional PIC in fading and AWGN environments also with severe near far problem. For examples, with 2 stage of interference cancellation, the proposed BA-PIC showed near single user performance for the weakest user under high system load of K/N ≈ 0.65 and power of interferers as high as 15dB.

The MAI estimation performance and achievable spectral efficiency of the proposed blind adaptive PIC and SIC techniques in Rayleigh fading environments is also investigated and com- pared with other receivers. The spectral efficiency in bps/Hz is calculated based on output SINR, which is also the function of MSE of MAI estimation conditioned on users’ data. Particularly, the blind PIC is shown to give highest ≈ 7.3 bps/Hz compared with ≈ 5.3 for conventional PIC, ≈ 4.5 for CMA-SIC, ≈ 3.0 for SIC, and ≈ 2.7 for MF receivers.

In Section 3.4, a low complexity subcarrier combining technique for MC-CDMA receiver is presented in mobile Rayleigh channel. It is shown that, by exploiting the structure formed by users’ repeating sequences into an adaptive algorithm with simple CM cost function and judicious selection of step-size, simultaneous supersession of MAI and implicit channel tracking is achieved without any knowledge of the channel. Improved BER and user capacity compared with MRC and EGC are shown. For example, at the BER of 0.01, it can support 17 users compared with 7 and 4 users for MRC and EGC, respectively. Optimum step-size are also investigated using simulations to give more improved performance.

To sum up, this Chapter has made an effort to apply a blind adaptive approach to tackle the MAI and imperfect channel estimation problems in the uplink of CDMA. The next Chapter is focused to address the problems in achieving spatial diversity for the uplink of CCMA and CDMA. We propose to solve the problems of deep fade as well as the MAI in a joint manner by combining user collaboration and multiuser detection.

Chapter 4

Collaborative Space Diversity Schemes

4.1

Introduction

The work presented in this Chapter are aimed at finding new ways of providing spatial diversity in uplink of wireless systems employing multiple access schemes such as CCMA and CDMA. These investigations are inspired by popular techniques such as space-time coding [73, 2], cooperative diversity [8, 7], which are used extensively for single user systems to improve the outage capacity and to counter the effects of slow fading, shadowing etc. There has been relatively very little work on applying these techniques to solve the problem of degraded BER performance under multiuser settings. The main reasons behind the lack of research in this subject area are:

• The uplink of multiuser systems naturally comes with problem of reduced orthogonality (in the case of CDMA) among the users’ transmissions and hence the loss in achievable rate due to MAI

• The interference problem arising in such environments requires more advanced MUD re- ceivers, which may complicate the performance analysis significantly.

The contributions presented in this Chapter attempts to fill this gap to some extent. It is how- ever acknowledged that there is a need for more research work to make full use of the user co- operation diversity technique in practical multiuser wireless systems. The term cooperative diver- sity is usually used to denote the spatial diversity that is achieved via cooperation of users/nodes [7, 8, 79, 75, 85], which is also referred to here as collaborative diversity. Although there are dif- ferences where the word Collaboration is used, the two terms are treated equivalent in this Chapter and hence used interchangeably.

The first two work are proposed for uplink of CCMA to provide transmitter diversity gain. In the first scheme presented in Section 4.2, diversity gain is achieved by simply allowing each user node to transmit from one antenna at a time. The latter scheme described in Section 4.3, is focused on providing the diversity gain from user cooperation and hence presents a practical alternative to

the former. The cooperative diversity technique is also investigated for uplink of CDMA with MAI. Two schemes are presented for CDMA that address the issues of providing transmitter di- versity gains while minimizing the effects of the MAI. More specifically, new schemes consisting of user cooperation diversity and successive interference cancellation technique are proposed and evaluated in Section 4.4. Finally, in Section 4.5, a new bandwidth efficient collaborative diversity (BECD) scheme is proposed and evaluated in uplink of CDMA.