Spectrum Sharing Methods in Coexisting Wireless Networks

Radio spectrum, the fundamental basis for wireless communication, is a finite resource. The development of the expanding range of radio based devices and services in recent years makes the spectrum scarce and hence more costly under the paradigm of extensive regulation for licensing. However, with mature technologies and with their continuous improvements it becomes apparent that tight licensing might no longer be required for all wireless services. This is from where the concept of utilizing the unlicensed bands for wireless communication originates. As a promising step to reduce the substantial cost for radio spectrum, different wireless technology based networks are being deployed to operate in the same spectrum bands, particularly in the unlicensed bands, resulting in coexistence. However, uncoordinated coexistence often leads to cases where collocated wireless systems experience heavy mutual interference. Hence, the development of spectrum sharing rules to mitigate the interference among wireless systems is a significant challenge considering the uncoordinated, heterogeneous systems. The requirement of spectrum sharing rules is tremendously increasing on the one hand to fulfill the current and future demand for wireless communication by the users, and on the other hand, to utilize the spectrum efficiently. In this thesis, contributions are provided towards dynamic and cognitive spectrum sharing with focus on the medium access control (MAC) layer, for uncoordinated scenarios of homogeneous and heterogeneous wireless networks, in a micro scale level, highlighting the QoS support for the applications. This thesis proposes a generic and novel spectrum sharing method based on a hypothesis: The regular channel occupation by one system can support other systems to predict the spectrum opportunities reliably. These opportunities then can be utilized efficiently, resulting in a fair spectrum sharing as well as an improving aggregated performance compared to the case without having special treatment. The developed method, denoted as Regular Channel Access (RCA), is modeled for systems specified by the wireless local resp. metropolitan area network standards IEEE 802.11 resp. 802.16. In the modeling, both systems are explored according to their respective centrally controlled channel access mechanisms and the adapted models are evaluated through simulation and results analysis. The conceptual model of spectrum sharing based on the distributed channel access mechanism of the IEEE 802.11 system is provided as well. To make the RCA method adaptive, the following enabling techniques are developed and integrated in the design: a RSS-based (Received Signal Strength based) detection method for measuring the channel occupation, a pattern recognition based algorithm for system identification, statistical knowledge based estimation for traffic demand estimation and an inference engine for reconfiguration of resource allocation as a response to traffic dynamics. The advantage of the RCA method is demonstrated, in which each competing collocated system is configured to have a resource allocation based on the estimated traffic demand of the systems. The simulation and the analysis of the results show a significant improvement in aggregated throughput, mean delay and packet loss ratio, compared to the case where legacy wireless systems coexists. The results from adaptive RCA show its resilience characteristics in case of dynamic traffic. The maximum achievable throughput between collocated IEEE 802.11 systems applying RCA is provided by means of mathematical calculation. The results of this thesis provide the basis for the development of resource allocation methods for future wireless networks particularly emphasized to operate in current unlicensed bands and in future models of the Open Spectrum Alliance