In addition, the load for time synchronization
should be shared, so some nodes in the network do not fail earlier than
others. If some parts of the network fail earlier than others, the partitioned
networks may drift apart from each other and become unsynchronized.
??? Server-less: A precise time server may not be available. Even if the time
servers are available, they may fail when placed in the sensor field. As a
result, sensor nodes should be able to synchronize to a common time without
the precise time servers. This server-less feature also helps to address
the robustness challenge as stated earlier. Also, if precise time servers are
used by the server-less architecture, the quality of the synchronized clocks
as well as the time to synchronize the clocks of the network should be
much better.
??? Light-weight: The complexity of the time synchronization protocol has to
be low in order to be programmed into the sensor nodes. Besides being
energy limited, the sensor nodes are memory and CPU limited as well.
222
Chapter 9 Time-Synchronization Challenges and Techniques
The synchronization protocol may be programmed into a FPGA or designed
into an ASIC. By having the time synchronization protocol tightly
integrated with the hardware, the delay and variation of the processing
may be smaller. With the increase of precision, the cost of a sensor node
is higher.
??? Tunable service: Some services such as medium access may require time
synchronization to be always ON while others only need it when there
is an event.
Pages:
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366