Dynamic Resource Allocation in OFDM-based Cognitive Radio Networks

his dissertation addresses two important issues in cooperative cognitive radio networks: spectrum sensing in physical layer and dynamic resource allocation in MAC layer. The goal of spectrum sensing is to find the vacant spectrum bands opportunistically and have them accumulated in a spectrum pool. In the MAC layer the detected bands are allocated to the Secondary Users (SUs) dynamically to take the best advantage of the temporarily granted resources.;We first assume a general model of a multiuser OFDM cellular network with non-cognitive users. Then, a dynamic resource allocation scheme is proposed based on Nash Bargaining Solution (NBS) for the downlink of the network. Although NBS provides a fair and optimum approach in order to maximize the total rate of the network, there is no simple solution for the case of K users. Since NBS relates to maximizing the multiplication of the users utilities rather than the summation it makes the optimization problem harder to solve. Hence, by decomposing the NBS problem into two sub-problems, the power allocation reduces to the well-known water-filling algorithm and the subchannel assignment leads to a simple algorithm which takes the total channel gain of each user as the fairness factor. The proposed algorithm satisfies the minimum rate requirement of each user first and allocates the excess subchannels by searching over the K × N subchannels to noise ratio matrix based on the bargaining process happening between the base station and the users. Hence, the complexity reduces from order O( KN) to O(N × K). Simulation results show that the proposed algorithm keeps a balance between the Max-Min approach and the Max-Sum where Max-Min aims at maximizing the worst user rate and Max-Sum maximizes the sum of the rates by blocking the users in the poor channel conditions. We demonstrate the application of this approach to LTE-Advanced systems as well.;In the ensuing work, we extend our system model to a cognitive radio network where secondary users need to monitor the spectrum of the primary for possible idle bands. As the sensing is the first step for the secondary operation we investigate the available centralized and distributed spectrum sensing schemes. We propose a collaborative spectrum sensing method based on Stackelberg game in order to improve the sensing performance of cognitive radio users, especially when they are operating under severe channel fading. In this scheme the users with acceptable received SNR are allowed to lead the network sensing process and share their observations with the ones experiencing weak channel conditions. By defining a new and more appropriate metric to evaluate the performance of the collaborative detection, the proposed method addresses the drawbacks of the conventional centralized or distributed spectrum sensing. Moreover, it does not require exchange of channel information among nodes and only minimum reporting of local observations is needed. The simulation results indicate that the proposed scheme improves the network sensing performance and reduces the overhead as compared to the non-cooperative case and the conventional collaborative schemes, respectively.;In order to allocate resources in a cognitive OFDM based network the availability of subchannels are needed to be considered. Hence, the imperfect sensing information will reflect in resource allocation procedure. We extend our proposed dynamic subchannel and power allocation scheme based on Nash Bargaining Game for an ad hoc network of Secondary Users (SU) coexisting opportunistically with a Primary base station. Each SU is equipped with an energy detector to sense the Primary User (PU) activity over N OFDM sub-channels. We deploy Nash Bargaining Solution to model the power and subchannel allocation of the SUs over the temporarily available transmission opportunities. Simulation results show that the proposed dynamic power and subchannel assignment is simple, effective and fair. Moreover, the total throughput of the network is highly dependent on the accuracy level of the sensing information and the percentage of PU silence. An acceptable secondary network throughput is achievable if the probability of PU presence is limited to less than...