Wireless Ad Hoc and Sensor Networks

340 Wireless Ad Hoc and Sensor NetworksMoreover, resources such as bandwidth and buffer space at the sensornodes are quite limited. Fairness becomes a critical issue in the presenceof tight resource constraints and when several sensor nodes access theshared wireless channel in the CSMA/CA paradigm. For example, duringa forest fire application, not only sensors within a cluster collect a largeamount of data, but also it is desirable to collect the same amount of datafrom each deployed cluster, so that the temperature gradient can beinferred (Woo and Culler 2001). Therefore, a fair allocation of bandwidthfrom each node over multihops is required to transmit data to the BS.Thus, the scheduling protocol should deliver data from each sensor nodefairly during unusual events, and it also must incorporate the timevaryingchannel state to meet certain throughput and end-to-end delayrequirements.In this section, we introduce a novel sleep mode for CSMA-based asynchronousnetworks during idle periods, thus reducing energy spent overthe radio communication to a minimum. A cluster topology similar tothat of Heinzelman et al. (2002) is utilized with only periodic communicationbetween sensors and its CH. Thus, sensors are in sleep mode formost of the time, saving energy. However, in the case of an event, ameasuring circuit will be able to wake up the sensor to process and sendthe data to the CH. During sleep mode, because the sensors do not listenfor an incoming traffic, a query or maintenance packet sent from the CHto the sensor will not be received, which is clearly a drawback. However,the CHs buffer packets for the sensor and delivers them periodically inthe sensor wake state. Because sensors usually send data whereas onlysporadically receive packets or queries, the proposed sleep mode willgreatly increase energy savings while minimizing latency.A large amount of energy is also consumed during active periods whenthe nodes transmit data. In the case of an unusual event, the traffic generatedby the sensors can be huge. A significant percentage of energy willbe spent on transmission; hence, transmitter power control is importantto save energy. The DPC algorithm from Jagannathan et al. (2006) andfrom Zawodniok and Jagannathan (2004), for cellular and ad hoc networks,takes into account different radio channel uncertainties in a wirelessnetwork: path loss, shadowing, and Rayleigh fading. The DPCimplementation proposed in Zawodniok and Jagannathan (2004) wasdesigned for ad hoc, nonpersistent CSMA/CA wireless networks. In thissection, this MAC protocol utilizes a scheme for sleep mode in clustered,CSMA-based sensor networks, thus increasing the lifetime of sensor nodeswhile meeting the application constraints. Moreover, the proposed protocoluses the DPC scheme from the previous chapter to estimate the timevaryingchannel and to update the transmitter power.Finally, a DFS protocol from the previous section is incorporated withthe proposed energy-aware scheme so that certain QoS requirements can