SDMA performance and characteristics

Figure 3 SDMA spread spectrum system model

Figure 4 Single-user performance of SDMA system under AWGN channel

Figure 5 Performance of two-user asynchronous CDMA system based on different spreading codes under AWGN channel (PR-QMF spreading factor is 32, and length of Gold sequence and M sequence is 31)

In the previous analysis, it was mentioned that each channel of the SDMA system has different bit error rate and transmission rate. For the former, the bit error rate of each channel can be made close by power control; for the latter, the unfairness of this channel But it is inherent in the SDMA system. For users of the same service, let them use the same channel in a time-sharing manner to narrow the gap between the average transmission rates of each user. For example, in an odd time slot, user A uses sub-channel C1. B uses sub-channel C2; while even time slots, user A uses C2 and user B uses C1. For users with different services, this unfairness may be one of the advantages of SDMA systems: using high-rate channels to transmit broadband services ( Such as images), use low-rate channels to transmit narrow-band services (such as voice), and the system naturally combines these services together, so this will be suitable for transmitting multimedia signals.
In fact, the number of sequences in the spreading sequence family close to orthogonal (or quasi-orthogonal) is relatively small, so when the channel capacity is fixed, the capacity of the CDMA system using this spreading sequence is correspondingly small [ 9]. In the actual CDMA system, in order to increase the system capacity, a large number of non-orthogonal spreading sequences are usually used, but this will directly lead to the existence of near-far problem, which seriously affects the system performance. In the SDMA system, when the fundamental wavelet w is determined, the wavelet function family formed by it is theoretically orthogonal, and there are more spreading functions available (because it can be guaranteed from the frequency or time domain Its orthogonality), so it can better suppress the near-far effect, thereby reducing the complexity of the receiver.
The performance of the SDMA spread spectrum system is closely related to the selected fundamental wavelet w, which can be measured by analyzing the coding gain Gs of the filter bank. Here

(14)

Where σ2i is the variance of the output of the i-th analysis filter, and K is the number of layers of wavelet decomposition. The length of the synthesis / analysis filter is L (that is, the support length of the orthogonal wavelet, and it is even) and the given signal spectral density In the case of P (ω), the choice of the optimal wavelet that guarantees the best system performance comes down to In the closed concentration of the dimensional space, look for the best advantage to make Gs reach the maximum value. Figure 6 (a) and (b) respectively give some numerical calculation results of the signal spectral density with uniform distribution and truncated Laplace distribution.

(a) Evenly distributed,
(b) Truncated Laplace distribution, p (ω) = ec | ω | cos (ω / 2)
Figure 6 Local optimal results of Gs under different information spectral densities (L = 6)

In the numerical analysis in Figure 6, the local optimal search results are used instead of the global optimal values, because there is currently no perfect algorithm for solving the global optimal results. In addition, when the spectral density of the signal is more complicated, the search is the most The complexity of the optimal results also increases accordingly.