Resource and power management in next generation networks

The limits of today’s cellular communication systems are constantly being tested by the exponential increase in mobile data traffic, a trend which is poised to continue well into the next decade. Densification of cellular networks, by overlaying smaller cells, i.e., micro, pico and femtocells, over the traditional macrocell, is seen as an inevitable step in enabling future networks to support the expected increases in data rate demand. Next generation networks will most certainly be more heterogeneous as services will be offered via various types of points of access (PoAs). Indeed, besides the traditional macro base station, it is expected that users will also be able to access the network through a wide range of other PoAs: WiFi access points, remote radio-heads (RRHs), small cell (i.e., micro, pico and femto) base stations or even other users, when device-to-device (D2D) communications are supported, creating thus a multi-tiered network architecture. This approach is expected to enhance the capacity of current cellular networks, while patching up potential coverage gaps. However, since available radio resources will be fully shared, the inter-cell interference as well as the interference between the different tiers will pose a significant challenge. To avoid severe degradation of network performance, properly managing the interference is essential. In particular, techniques that mitigate interference such Inter Cell Interference Coordination (ICIC) and enhanced ICIC (eICIC) have been proposed in the literature to address the issue. In this thesis, we argue that interference may be also addressed during radio resource scheduling tasks, by enabling the network to make interference-aware resource allocation decisions. Carrier aggregation technology, which allows the simultaneous use of several component carriers, on the other hand, targets the lack of sufficiently large portions of frequency spectrum; a problem that severely limits the capacity of wireless networks. The aggregated carriers may, in general, belong to different frequency bands, and have different bandwidths, thus they also may have very different signal propagation characteristics. Integration of carrier aggregation in the network introduces additional tasks and further complicates interference management, but also opens up a range of possibilities for improving spectrum efficiency in addition to enhancing capacity, which we aim to exploit. In this thesis, we first look at the resource allocation in problem in dense multitiered networks with support for advanced features such as carrier aggregation and device-to-device communications. For two-tiered networks with D2D support, we propose a centralised, near optimal algorithm, based on dynamic programming principles, that allows a central scheduler to make interference and traffic-aware scheduling decisions, while taking into consideration the short-lived nature of D2D links. As the complexity of the central scheduler increases exponentially with the number of component carriers, we further propose a distributed heuristic algorithm to tackle the resource allocation problem in carrier aggregation enabled dense networks. We show that the solutions we propose perform significantly better than standard solutions adopted in cellular networks such as eICIC coupled with Proportional Fair scheduling, in several key metrics such as user throughput, timely delivery of content and spectrum and energy efficiency, while ensuring fairness for backward compatible devices. Next, we investigate the potentiality to enhance network performance by enabling the different nodes of the network to reduce and dynamically adjust the transmit power of the different carriers to mitigate interference. Considering that the different carriers may have different coverage areas, we propose to leverage this diversity, to obtain high-performing network configurations. Thus, we model the problem of carrier downlink transmit power setting, as a competitive game between teams of PoAs, which enables us to derive distributed dynamic power setting algorithms. Using these algorithms we reach stable configurations in the network, known as Nash equilibria, which we show perform significantly better than fixed power strategies coupled with eICIC.