Recent evolution of communication networks comprises of different segments and technologies, where each segment maybe implemented using different QoS. Further, the proposed all-IP core infrastructure of the future networks will offer varying QoS level multimedia services to the users. However, IP being a best effort service, seamless provision of end-to-end QoS guarantees is extremely important.
In today's world, devices with multiple networking capabilities is quite common. The traditional approach in networking includes grouping identical traffic and allocating them to the network that has the maximum available data rate. This creates unbalanced traffic load in the network, leading to poor utilization of the associated resources. This problem can be greatly alleviated if the traffic can be allocated intelligently to the networks. For fair traffic distribution, we modeled the AP of each network as a single queuing server. Then, suitable equations and algorithms are designed to divide the incoming traffic flow into multiple subflows and allocated to the APs based on their available data rates.
Network Selection in a Heterogeneous Cognitive Radio Wireless Network is a challenging task, since the users need to select the appropriate channels of the network in addition to the network itself. The varying levels of interference experienced by the secondary user (SUs) is due to the presence of primary user (PU)s in the adjacent channels. Hence, SUs transmitting highly sensitive data must find a channel that is interference free. In this dissertation, we develop a novel network and channel selection scheme that categorizes both the user applications and the network channels depending on their sensitivity level for interference and select them using a bipartite graph matching algorithm.
The effectiveness of Cognitive Radios is based on opportunistic access of the licensed channels by SUs while protecting the PU transmission. But channel sensing incurs cost in terms of time overhead and energy consumption. However, infrequent sensing also results in loss of transmission opportunity for the SUs. Hence, an interesting and challenging question arise: when should the SU sense the channel, sleep or transmit, to minimize the total cost? In this dissertation, we developed a novel scheme for deriving the optimal inter-sensing duration in a Cognitive Radio network, on the requirement of protecting the PUs' communications while minimizing the cost for the SUs. The scheme has been presented for both non-erroneous and erroneous channel sensing conditions.
Handling the "mobile data tsunami" in the future and providing indoor coverage is a significant challenge for the operators. The answer is LTE femtocells. However, limited spectrum availability in the cellular networks causes severe interference in the neighboring femtocell users that are transmitting in the same radio band. In densely deployed environments, interference problems in co-channel femtocells causes significant degradation in performance. In this dissertation, we proposed a CASFR scheme, that assigns distinct set of PRBs to each interfering femtocells in the downlink. In the uplink we proposed a PSE algorithm to further reduce any interference that may remain after performing CASFR. Finally, the topics for future work have been clearly identified.