Since this could take as long as one
period, which can be very long, data forwarding might su?®er substantially.
This could be important depending on the frequency of such wakeup events
and on whether it is more likely that a wakeup is needed in an already active
part of the network. If, on the other hand, data has priority over wakeup,
no delay bound can be guaranteed as the wakeup event could be queued
behind ongoing data transmissions [28]. To completely avoid such interference
and arrive at an out-of-band wakeup solution, di?®erent channels have to be
used, for example implemented as di?®erent frequency bands. If in addition,
this frequency separation is implemented by providing a separate radio for
data transmission (governed by the MAC) and wakeup, one node can be
engaged in tra?±c forwarding and at the same time participate in wakeup [13].
This complete decoupling of the two functionalities also has the benefit that
wakeup could potentially use another type of radio, which allows combining
the benefits of asynchronous wakeup with those of paging. However, the range
disparity problem would also be inherited in this case.
6 Conclusions
Three distinct classes of wakeup solutions (paging, synchronous and asynchronous
wakeup) were discussed, where each class has its own benefits and
drawbacks. In the current state of research, there is no clear best algorithm,
and it is expected that there will be no single solution that best fits all possible
scenarios.
Pages:
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354