Utility Aware Resource Management for Device to Device Communication

Abstract

Since the deployment of 1G cellular networks, cellular operators are constantly striving towards achieving higher data rates. Next generation networks are expected to meet the demands of subscribers for high data rates. To meet these challenging requests of higher data rates, the telecom research community is focusing its attention towards hexa dimensional resource management techniques in next generation cellular networks. These techniques include use of heterogeneous networks (hetnets), power allocation, channel allocation, broad band and efficient spectrum utilization, noise control methods and interference management. In next generation networks, a major portion of system throughput enhancements is expected to come from device to device (D2D) communication which is viewed as part of hetnets. In this thesis, D2D communication is considered the main enabling feature of hetnets to enhance throughput and its intelligent resource management is imperative. The scope of this work is further broadened by integration of cognitive radio network (CRN) concepts into the D2D resource management. To further boost data rates, D2D concept is extended by incorporation of communication among devices to devices (Ds2Ds) and their intelligent resource management. Resource management is a domain wherein resources are intelligently allocated to various users. Resource management aspects considered in this thesis include admission control, mode selection and power allocation. This work considers D2D joint resource allocation to maximize the overall throughput of cellular networks having D2D capabilities while observing the sanctity of constraints of power, quality of service (QoS) and interference. This problem is NP-complete and is classified as mixed integer non-linear optimization problem. Computational complexity of the exhaustive search algorithm (ESA) increases exponentially with the number of users. This makes it very difficult to find optimal solution for such kind of problems. In this work, convergent solutions for solving the joint resource allocation problem in D2D communication has been presented. By exploiting the special structure of the problem, outer approximation algorithm (OAA) and mesh adaptive direct search (MADS) algorithms have been used to obtain epsilon-optimal solution. Using joint utility function, different scenarios of simulation results confirm effectiveness of MADS viz-a-viz OAA. Performance of MADS was found to be equal to that of ESA. Cognitive assisted D2D network is an emerging domain which utilizes television (TV) white space spectrum. Joint resource allocation of CRN assisted D2D networks along with constraints of transmit power, QoS and interference (related to TV users) is considered in this thesis. The problem is mathematically modelled with a utility function having two terms related to maximization of admitted users and maximization of overall system throughput. MADS algorithm is adopted for solution of NP-complete problems. Simulation results compare different scenarios using ESA and MADS algorithm. Results of simulation testify to the fact that MADS provides solution close to optimal with reasonable complexity. In Ds2Ds resource management problem, the objective is to maximize energy efficiency of Ds2Ds communication, subject to compliance of constraints related to radio resources, quality of service (QoS) and interference. Performance of Ds2Ds is bench marked with multi-homing technique and devices to device (Ds2D) communication. The objective of maximization of energy efficiency reinforces the much sought-after green communication.