7.7 Examples
Previously, we saw that eigenvector beamforming coupled with a simple MMSE-based
receiver can outperform spatial spreading even when an iterative receiver is employed
together with the latter. In this section, we quantify the computational complexity
associated with these two approaches [12]. Specifically, we consider the 2?—2 system
operating at 39 Mbps introduced in Section 9.6.3. The number of information bits is Nb =
MIMO Spatial Processing for 802.11n WLAN 214
8000. Table 9.4 shows the equivalent number of real MUL and ADD operations for the
following four configurations:
1. Spatial spreading; 2 streams of QPSK, R = 3/4; NO = 52; iterative LSD-SOVA receiver
with Nit = 1 iteration (Nave(1) = 2.20; Nave(2) = 2.33).
2. Same as 1., but with Nit = 2 iterations.
3. Eigenvector beamforming; 1 stream of 16-QAM, R = 3/4; NO = 52; MMSE receiver.
4. Eigenvector beamforming; 2 streams, 16-QAM, QPSK, R = 1/2; NO = 52; MMSE
receiver.
Table 9.4: Complexity comparison of iterative
LSD and MMSE-based receivers.
Real MUL Real ADD
1. 40.12 440.90
2. 45.25 1122.02
3. 19.86 453.66
4. 27.95 465.30
Note that the fourth configuration employs different modulations on the two streams.
Pages:
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513