On Interference Management With Incomplete Side Information

Sammanfattning: The evolution towards next generation wireless networks has to serve high expectations and demands, therefore ongoing research has to tackle problems that have hitherto received less attention. A decisive step in this direction is to consolidate cohabitation paradigms into the envisioned communication protocols. There hasbeen increasing interest in enabling, possibly heterogeneous, entities to coexist harmoniously;recent studies have established theoretical gains that stem from coordinated interference embedding into various network architectures. To further bridgethe gap between theory and practice we need to explore uncharted dimensions of interference networks and, in particular, the limitations that arise in realistic scenarios. In this thesis, we consider the performance optimization and analysis of interference networks where the participating links have equal or unequal quality-of-service priorities. Link priorities are widely applied in the context of cognitive radio systems and are reflected in the metrics and the relative weights that are assigned to each link in the network. To integrate conditions that are encountered in practical systems we assume scenarios where the link of interest does not know perfectly the relevant design parameters, namely the channels, but instead has an estimate. We propose and analyze metrics that are suitable proxies for performance assessment in the considered scenarios and as a case-study we establish a framework for enhanced parameter acquisition.In the first part of this thesis, comprising the first three technical chapters, we focus on single-user system design within the interference network. In these chapters, the main goal is to optimize the design or analyze the performance, or even carry out both tasks, for the target link. We consider metrics that are widely adopted for performance assessment in wireless communications such as the instantaneous achievable rate (Chapter 2), its ergodic counterpart (Chapter 3), or the probability of decoding outage (Chapter 4). Our setups are fortuitous in that they have direct application to the well-known underlay cognitive radio framework. However, we provision for a wider context by occasionally casting the discussion at the more general level of cellular networks, as in Chapter 3. The presence of other communication links is reflected in the constraints that bound the outgoing interference power of the desired system, in each respective scenario, towards the unintended nodes. Limiting the inflicted interference is a way of mapping the remaining participants’ quality-of-service and subsequently their priority in the network. To capture realistic conditions, we pursue the aforementioned design targets under the assumption that the system of interest has incomplete knowledge of the channels that are involved in the problem formulation, namely in the objective and the constraints. One of our contributions is that we propose formulations that are tailored to the type of the available channel side-information and suitable for the, pertinent to the model, performance assessment. In several cases, we provide novel analytical solutions to the associated optimization problems and establish insights that naturally extend existing results into our scenarios.In the second thematic part, which maps to Chapter 5 of the thesis, we generalize our investigation to the joint design of multi-user interference networks. In this chapter, we seek to optimize the performance of all links simultaneously and in away that is beneficial for all the participants in the network. Under this premise, trade-offs inevitably arise and sets of solutions that are Pareto-optimal become the desirable goal. Even though this area is well-documented, the assumption of incomplete channel state information introduces a new dimension into the analysis; to settle the arising shortcoming we stipulate and tailor to our setup the definition of the outage rate-region, which essentially is a blend of the metrics used in the previous chapters. Thus the theoretical findings and results presented in this chapter generalize gracefully the content of the preceding part, which can be obtained as a snapshot of the generic model of Chapter 5. The analysis conducted in this chapter, along with the conclusions we draw, indicate that we can successfully replace orthogonal transmission schemes with cohabitation protocols.Throughout the thesis we appeal to the availability of incomplete channel-state information to render our models more realistic. However, the acquisition of any necessary exterior parameters is not explicitly accounted for, in the first two parts. To fill this gap, we present in Chapter 6 a short case-study, which is applicable to some of the setups considered in the thesis. In particular, we investigate enhancement techniques for improving the estimation of the instantaneous signal-to-noise ratio, when viewed as a deterministic parameter. Albeit its narrow scope, this study brings forth some interesting ideas for enhancing the quality of the available side-information.