DEVELOPMENT OF A SEQUENTIAL QUADRATIC PROGRAMMING BASED ENERGY EFFICIENT COGNITIVE RADIO NETWORK

Abstract

Huge increase in wireless devices demands e cient utilization of frequency spectrum. Cognitive radio is an e cacious technology to increase the spectral e ciency. The idea is to exploit the unused frequency bands more e ciently. The spectrum is cooperatively sensed by the secondary users (SUs) to detect the occupancy of the frequency bands and opportunistically access the available vacant bands. Since multiple secondary users are interested in transmitting data, code-division multiple access (CDMA) is used to avoid interference among di erent secondary users. The power allocation issue in CDMA-based cognitive radio networks has been examined in this thesis for SUs through a two-phase protocol, namely spectrum sensing and data transmission. In rst phase, availability of frequency bands is detected through cooperative spectrum sensing using OR rule. In the second phase, data is transmitted by the SUs using CDMA scheme if the band is estimated to be free. The sensing energy is an increasing function of false-alarm probability whereas transmission energy is a convex function of false-alarm probability. Therefore, it is di cult to nd optimal false-alarm for simultaneously optimizing sensing and transmission energies. Both the sensing energy and transmission energy could be optimized separately or jointly using copolyblock algorithm. In separate optimization method, secondary users are allocated transmit power irrespective of the sensing parameters. The constraints on the detection probability and data rate of secondary users are taken into account during optimization. It has been observed from literature that joint optimization saves more energy than it's separate counterpart for low sensing signal-to-noise ratios (SNRs) and high data rates. However, joint optimization using copolyblock algorithm su ers from few disadvantages such as high storage capacity, high computational complexity and it also does not guarantee convergence. To overcome these limitations, sequential quadratic programming (SQP) based energy optimization has been proposed in this thesis. Simulation results show that for low sensing SNRs, joint optimization based on SQP leads to more energy saving as counterpart based on copolyblock algorithm. Furthermore, energy saving increases with increase in data rate. Further analysis has revealed that the MAJORITY rule based cooperative spectrum sensing (CSS) scheme performs better than that of the OR rule. Hence, this thesis proposes that SQP-based MAJORITY rule is a better candidate for energy optimization in CSS