Wireless Ad Hoc and Sensor Networks

178 Wireless Ad Hoc and Sensor Networkschannel fluctuations. In this chapter, first, a linear system theory isemployed to analytically represent the interference problem among usersin a cellular network. Note that state space modeling and controller designis a preferred method in other application areas as well. State space andoptimal schemes (Jagannathan et al. 2002, Dontula and Jagannathan 2004)are first developed to control the power of each transmitter in the presenceof path loss uncertainty. Subsequently, a channel state estimator embeddedin the power control scheme (Jagannathan et al. 2006) is presentedso that the receiver can suggest suitable transmitter powers required toovercome slowly varying channel conditions.5.1 IntroductionControl of transmitter power allows communication links to be establishedin a channel, using minimum power to achieve required signal-to-interferenceratios (SIR) reflecting QoS levels. Interference mitigation properly andquickly increases network capacity through channel reuse. Past literatureon power control (Aein 1973, Alavi and Nettleton 1982) focused on balancingthe SIRs on all radio links using centralized control. Later, distributedSIR-balancing schemes (Zander 1992, Grandhi et al. 1994) weredeveloped to maintain QoS requirements of each link. In a dynamic environment,maintaining and guaranteeing the QoS is extremely difficult,because maintaining QoS of all existing links may require removal of someactive links, selectively admitting new links, and/or providing differentiatedQoS services. Power control protocols were addressed in the literaturefor cellular networks initially and later extended to ad hoc networks.Mitra (1993) proposed a distributed asynchronous online power controlscheme, which can incorporate user-specific SIR requirements and yieldminimal transmitter powers while converging geometrically. However, asnew users try to access the channel, the SIRs of all existing ones may dropbelow the threshold, causing an inadvertent drop of ongoing calls (Hanly1996). A second-order power control scheme that uses both current andpast values to determine the transmitter power for the next packet, wasproposed in Jantti and Kim (2000). This scheme, developed by applying thesuccessive overrelaxation method, converges faster, but the performance ofthe power control scheme can be affected negatively because of measurementerrors and loop delays. In the seminal work of Bambos et al. (2000),a DPC scheme is presented with admission control. An optimum transmitterpower control scheme similar to Bambos et al. (2000) with heterogeneousSIR thresholds was proposed in Wu (2000). The DPC scheme was able toincorporate inactive links with active link-protection mechanisms. The