Wireless Ad Hoc and Sensor Networks

258 Wireless Ad Hoc and Sensor Networksthe past transmission history, as in ARF protocol, or assuming that thechannel state does not change significantly between consecutive transmissions,as in the case of RBAR. Hence, the selected rate was often notoptimal. Additionally, these protocols neither take energy-efficiency intoconsideration nor modify the transmission power to save energy duringrate adaptation.The operation of the ARF protocol (Kamerman and Monteban 1997) ispresented in Figure 6.12. The four rates (R1, R2, R3, and R4) are consideredwith corresponding SNR thresholds. In this example, the first packet is sentwith the maximum rate allowed for the channel state. The following packetsare sent using the same rate, though the SNR could have increased thuslowering the throughput for the current channel state. After three consecutive,successfully received packets, the rate is then increased even though thechannel state could have changed significantly during this time. The newrate (R2) used for the fourth packet could be still lower than the maximumpossible rate. On the other hand, when the SNR decreases, the selected ratewill always be higher than the acceptable throughput possible, resulting inproblems of decoding of packets at the receiver.In short, the problems observed in the ARF protocol are the result of thelack of information about the radio channel state, because, no measurementsof signal reception are considered. In consequence, the throughputachieved by ARF is lower than possible for a given channel state. Furthermore,the energy is consumed inefficiently.SNR(t)Packet OK with optimal ratePacket OK but with sub-optimal ratePacket LOST because min. SNR not metR4R3R2R1R1 R1 R1 R2 R2 R2 R3 R3 R3 R4 R4 R4 R33 successful = > 3 successful = > 3 successful = > 3 failures = >increase rate increase rate increase rate decrease rateFIGURE 6.12ARF rate selection.