Power Control And Multi-target Identification In Cognitive Wireless Networks
Le, Hong-Sam Thi
MetadataShow full item record
Recent research results have shown that cognitive wireless networks (CWN) have the potential to alleviate spectrum scarcity problem resulting from current policies for radio resource allocation management and dramatically improve the overall performance of communication systems. Unlike conventional wireless networks, which lack the flexibility and adaptation in their operations, CWNs exploit cognitive radio technology which provides cognitive wireless devices with the ability to sense the situation, adapt to the environment and take appropriate actions correspondingly. There have been many challenges in building a fully functional CWN. New developments and approaches need to be proposed to allow radio users to share primary licensed spectrum without harming the primary users. This thesis aims to address the problem of transmit power control in the scenario of spectrum sharing. We design a transmit power control system using Fuzzy Logic System to provide cognitive radios with the ability to coexist with primary (licensed) users in the same frequency band. With the built-in fuzzy power controller, a cognitive radio is able to opportunistically adjust its transmit power in response to the changes of the interference level to primary user (PU), the distance to PU and its received power difference at the base station (BS) while satisfying the requirement of suffciently low interference to PU. We increase the reliability of our power control scheme by using linguistic knowledge of transmit power control (TPC) obtained from a group of network experts. The outcome of this study show that our proposed fuzzy power control scheme leads to significant performance improvement in average transmit power consumption and average outage probability compared with the fixed-step power control scheme. A new application of CWNs may be found in radar sensor networks wherein cognitive radios act as cognitive radars. Our key purpose is to deal with multiple targets within a required surveillance region in a robust and cost-effective manner. Thus, we focus on the problem of jointly classifying and identifying multiple targets in radar sensor networks where the maximum number of categories and the maximum number of targets in each category are obtained a priori based on statistical data. The actual number of targets in each category and the actual number of target categories being present at any given time are assumed unknown. We then propose a joint multi-target identification and classification (JMIC) algorithm for radar surveillance using cognitive radars. The existing target categories are first classified and the targets in each category are then identified. We also show that the proposed JMIC algorithm is a well-suited approach to surveillance activities in the future cognitive radar sensor networks.