It is well known, as shown with WMSA, that using channel
estimates of not only the present and past but also the future can significantly improve performance.
Here we briefly compare two RAKE finger architectures and discuss the advantages and
disadvantages of both approaches.
The first approach is based on storing despread symbols; we will call this symbol level (SL)
buffering. A simplified block diagram of a RAKE finger is given in Fig. 7.78.
Let us assume we need to store 2 time slots of symbols in this buffer, which is equivalent to
(2 2560)/SF complex symbols. If we let the number of fingers equal Nfing and assume the I/Q
samples require 8 bits of finite precision representation, then the total number of bytes required is
given as
[bytes] (7.111)
The second approach is based on storing predespread oversampled chips; we will call this chip
level (CL) buffering. A simplified block diagram of a RAKE finger is given in Fig. 7.79.
SL
e c2*2560 chips
SF d*2I/Q*8 bits f
8
*Nfing
408 CHAPTER SEVEN
FIGURE 7.79 RAKE finger architecture based on chip-level buffering.
X
X
CCCH
X
CDCH
Finger
Input
*
Chip
Level
Buffering
MA
Finger
Output
Chips Symbols
SF
??‘
CCH
??‘
The advantage of this architecture is that the memory doesn??™t need to be repeated for each finger.
Pages:
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672