Hence this delay buffer allows us to have more accurate
channel estimates to be used in the time slot. The length of the FIFO is also a measure of the
degree of noncausality used in the receiver. The second block uses the past, present, and future estimates
to create a time varying channel estimate. Here the estimate is updated every symbol position
in order to track the time variations.
In Fig. 4.14 we show some more details of the interpolation, where we use four channel estimates
to aid the coherent detector. These four channel estimates, h1, h2, h3, and h4, are updated every time slot.
ro(t) > d(t)
rd (t) # r*p
(t) d(t) # Zh^(t)Z2 nh(t)
rd (t) # r*p
(t) d(t) # h(t) # h^*(t) n(t) # h^*(t)
rd (t) d(t) # h(t) n(t)
rp(t) > h^(t)
rp(t) h(t) np(t)
rp(t) ZpZ2 # h np(t)
rp(t) r (t) # p* p # p* # h(t) n(t) # p*
r(t) p # h(t) n(t)
MODULATION DETECTION TECHNIQUES 183
184 CHAPTER FOUR
Data P Data P Data P Data P
hn(t1) hn??“1(t1) hn??“2(t1) hn??“3(t1)
1.5 Time Slot
Buffer
Decision
Device
Data
d(t)
Detected
Symbols
( )*
a0(t) a1(t) a2(t) a3(t)
FIGURE 4.14 Noncausal channel estimation example.
They all enter the interpolation block that has time varying weights: a0(t), a1(t), a2(t), and a3(t), where
they are updated every symbol time, t kTs.
Pages:
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312