11n WLAN 212
9.7.4 LSD Max-log-MAP
On the average, only hypotheses for s are considered when computing the bit
LLRs, and only ??‘ operations are performed. The corresponding
upper bounds on these quantities are given by and , respectively. The
resulting average complexity, expressed as the equivalent number of real MUL and ADD
operations, is summarized in Table 9.3, where the indicator function 1full is 1 only in the
first iteration (otherwise 0), and 1it is 1 if a-priori LLRs are available, as they would
normally be in the second and subsequent iterations.
1 ( ) S N
ave i N i = ??‘
1 ( ) 2 S N
ave i N i = ??’ {} max ?‹…
S N
cand N 2 S N
cand N ??’
9.7.5 Per-Stream LLR Computation
When performing per-stream LLR computation as in the suboptimal MMSE-based
receiver, the number of hypotheses that need to be considered is reduced from
(for full-complexity max-log-MAP) to ??‘ . Further complexity reduction is made
possible by the fact that the cost functions take scalar inputs rather than vectors and
matrices. The complexity of per-stream max-log-MAP LLR computation is summarized in
Table 9.3.
1 2
NS
i i K = ??‘
1 2 S i N K
i=
9.
Pages:
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511