Wireless Ad Hoc and Sensor Networks

Distributed Power Control and Rate Adaptation 265TABLE 6.1Power Thresholds for Selecting the MostEnergy-Efficient RateSNR[dB]SNRRate[Mbps]Upper Thresholds ofTx Power [mW]10 10.00 1 > 20.0914 25.12 2 20.0922 158.49 4 10.6428 630.96 6 1.4932 1584.89 8 0.30REMARK 3The power threshold given by Equation 6.33 depends upon DPC calculationsfor the lowest supported rate. The power thresholds can be calculatedin advance, reducing online calculation overhead of the protocol.Table 6.1 presents the thresholds calculated for a set of five rates used inthe simulations. For example, if the transmission power calculated by theDPC algorithm is equal to 15 mW, 2 Mbps is the most energy efficientrate, because 15 mW belongs to a set defined between 20.09 mW (2 Mb)and 10.64 mW (4 Mbps).6.8.5 Modulation Rate to Overcome CongestionIn the previous subsections, the maximum throughput and energyefficientrates were calculated. The throughput increases with rate; however,the lower the rate used, larger the number of bits that can be transmittedper joule. Therefore, the rate adaptation has to generate a trade-off betweenthe throughput and energy-efficiency. Hence, the selection of a modulationscheme that results in the best throughput while satisfying the energyconstraint is discussed next. Moreover, in the proposed protocol, the bufferutilization at the transmitter is used as an indication of the requiredthroughput, and will impact the rate selection method.Because a high transmission rate implies high buffer occupancy at thereceiving node, provided the channel is conducive, the buffer occupancyat each node should be taken into consideration during rate adaptation.The basic idea is presented in Figure 6.16. Here the lowest supported rateis selected when the queue at the transmitting node has low utilization.Increased queue occupancy at this node indicates a higher traffic demand.Hence, the modulation rate that results in higher throughput is selected.This rate adaptation based on buffer occupancy can be explained as follows.The lowest supported rate will be chosen if the congestion is low. As thecongestion in the network increases, the queue occupancy will increase. Consequently,a higher rate will be selected to clear the congestion, rendering a