In addition, the control
and energy overhead of the switching itself has been neglected, as is commonly
done. However, some work has investigated the e?®ect of the switching overhead
as well [19] [11].
The above assumptions are typically reasonable first-cut approximations,
as can be seen in Figure 1. In this case, the average energy consumption,
normalized versus a scenario where a node never transitions to the sleep mode,
is given by:
Enorm ??
Tperiod
Tactive
. (7)
By combining equations (6) and (7), the normalized energy consumption
in the quasi-dormant state can be expressed as a function of the normalized
worst case wakeup delay:
Enorm ??
1
Dnorm + 1
. (8)
This equation expresses the fundamental tradeo?® between energy and delay
that is present in the majority of proposed wakeup solutions. Figure 4
illustrates the nature of this tradeo?®. By allowing a small delay, large energy
savings can be obtained initially. The exact behavior and shape of the tradeo?®
curve depend on the specifics of protocol [13]. Instead of analyzing various protocols
in detail, this fundamental tradeo?® provides a solid and easy method
for a rough comparison between them. Essentially, the tradeo?® behavior of
equation (8) underlies most of them, and the main di?®erence between them
is the minimum value of Tactive, which maps the normalized wakeup delay
to the absolute one (which is the actual constraint). Solutions with a smaller
Tactive are generally preferable, although other factors also come into play, as
will be detailed further in the later sections of this chapter.
Pages:
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332