In this thesis work, we focus on the design of a new channel allocation mechanism for the Cognitive Radio based xG networks, based on graph theory. We propose division of the users as well as the channels on the basis of the quality of service constraints. We then analyze the performance of our mechanism incrementally in two steps: (a) Channel allocation in a static environment. (b) Channel allocation in a dynamic environment. In the first scenario, we assume a static channel configuration, wherein the primary users' behavior has been strictly defined according to a schedule. We analyze the performance of our scheme in such an environment taking into account the various performance measures like packet drop rate. We compare our scheme against the bare bones scheme where no priorities are taken into account while allocating channels.
In the second part of our work, we simulate the performance of our algorithm in a dynamic environment, where the behavior of the primary users is arbitrary. In such a scenario, the algorithm must adapt to take into account the changing channel qualities which directly impact the users. We again compare our scheme based on graph theory, against the bare bones scheme where the priorities aren't taken into account as well as against the greedy scheme which simply serves the users based upon their priorities. We show that our approach delivers vast improvements in packet drop rates compared to the random channel allocation scheme.