Interference Coordination for Low-cost Indoor Wireless Systems in Shared Spectrum

Abstract: Mobile broadband services have become a big success over the last several years. Innovative, smart handsets have caused explosive traffic growth which has led to a severe capacity shortage. Since the majority of traffic originates from indoor locations or hotspots, significant invest- ment in indoor wireless infrastructure is predicted in order to resolve the capacity problem. While existing public operators mainly focus on high-mobility wide-area services, non-traditional local access providers (LAPs) such as facility owners are more and more interested in high data rate indoor services for their employees or customers. An obstacle is that they do not have access to dedicated spectrum. One proposal is spectrum sharing between LAPs. In shared spectrum, interference management emerges as a key technical challenge, and this becomes more critical as indoor systems become increasingly dense.This thesis concentrates on the interference management problem when spectrum is shared between high-density indoor wireless systems. There are two different design directions which require different system architectures. A Wi-Fi or femtocell system works in a fully uncoordinated manner without any inter-cell signaling. This allows high network scalability with cheap devices but leads to poor performance. Alter- natively, advanced interference coordination can be used. It certainly improves the performance; however, it usually requires expensive infras- tructure for real-time information exchange. A key question asked in this thesis is if the interference coordination gives sufficient economic gain to a LAP in terms of a total deployment cost. In order to answer this question, we first develop a conceptual framework to define and compare various levels of coordination. Then, we measure the re- quired number of access points (APs) at a given area capacity demand to estimate the economic gain.The coordination decision problem for a LAP is divided into two. Firstly, the LAP needs to choose the right level of coordination within its own network. Secondly, it determines whether or not to cooperate with neighboring LAPs for coordinating interference across the net- works. Regarding the intra-network decision, the comparison ranges from uncoordinated CSMA/CA to ideal interference cancellation. We find the total deployment cost of the uncoordinated CSMA/CA network soars when an area capacity requirement exceeds a certain threshold. The performance gain of the ideal coordination does not pay off the cost of high-speed backhaul because walls effectively suppress interference. Therefore, the most viable approach in a typical indoor environment is using dynamic coordination schemes via existing backhauls, for example Ethernet or xDSL. As for the cooperation decision, our major finding is that non-cooperative spectrum sharing is feasible provided that the transmit power of the APs is properly regulated. Although cooperation with advanced inter-network coordination schemes brings about cost savings, it is not sufficient to overcome practical barriers to a cooperation agreement especially when the capacity demand is high.