Caching and Network Coding for Smart Radio Access

The current generation has undoubtedly been witnessing a revolution of wireless network technologies in recent decades. First of all, there has been an enormous growth of wireless user equipments and network infrastructures in terms of both number and variety. Moreover, data traffic over wireless networks has significantly increased, especially since mobile video streaming and social networking has come into play. The demand is expected to further increase in the upcoming years such that current network infrastructures will not be able to support this demand. To tackle these problems and maintain uninterrupted, fast and reliable service, many different mobile radio access standards have been proposed and there have been many ongoing discussions on which of these numerous proposals should officially be standardized. Overall, the scope of this thesis is the development and the mathematical analysis of new techniques for mobile radio access that offer improved capacity, data rates and reliability. Particularly, we explore two of the key enablers of the future wireless standards by investigating the application of caching and network coding concepts to the evolution of the LTE standard. In the first part, we investigate caching. One of the main bottlenecks in future cellular networks is formed by the backhaul and a promising means of reducing backhaul traffic is by reserving some storage capacity at base stations and use these as caches. In this way, part of the data is stored at the wireless edge and the backhaul is used only to refresh this stored data. Data replacement will depend on the users’ demand distribution over time, which is varying slowly, and the stored data can be refreshed at off-peak times. Thus, caches containing popular content can serve users without incurring an additional load on the backhaul. In order to show the effectiveness of caching techniques, we tackle the problem from different perspectives. Firstly, we use tools from stochastic geometry and mathematical optimization to model and analyze the optimal geographical caching in heterogeneous networks, in which an operator aims to jointly optimize the cached content in base stations with different storage capacities for the case where the content popularities are fixed and known. Secondly, in addition to the aforementioned tools, we also use tools from game theory and combinatorics and further exploit the topological properties of the network, again for the case where the content popularities are fixed and known; and we provide a distributed asynchronous algorithm for optimizing the content placement which can be interpreted as giving the best response dynamics in a potential game. Thirdly, we focus on the case where the content popularities evolve over time. We use the tools from estimation theory in addition to aforementioned ones. We provide a simple estimator that estimates the content popularities dynamically and propose an algorithm that uses these estimated content popularities to optimize the content placement in base stations. Particularly, in the first part of the thesis, we mainly focus on maximizing the probability that a user will find the content that she requests in one of the caches that she is covered by. In the second part, we investigate network coding. Network coding is a family of coding technologies that promise more effective use of network resources. Many techniques have been proposed for broadcasting/multicasting data, serving multiple users by coordinated multipoint transmission elements and techniques that enable device-to-device (D2D) communication for future wireless standards. In all these techniques, the main obstacle is the requirement for perfect synchronization between the service elements and the user equipments in order to exploit the benefits endowed by these techniques that require coordination. In this thesis, we show that network coding is an effective tool to deal with uncertainty in wireless networks and alleviate the need for synchronization between network elements. We tackle the problem from both theoretical and practical perspectives. Firstly, we use the tools from queueing theory and analyze the decoding delay of a network containing a single source transmitting network coded packets via multiple routers to a receiver. Secondly, we develop a method for D2D assisted multicast service delivery to multiple equipments. Thirdly, we present two methods where network coding is used to mitigate the communication overhead between cooperating base stations. Finally, we propose an efficient repair scheme by combining network coding with broadcasting capability of the base stations. In a nutshell, the mathematical methods, results, analysis and the techniques presented in this thesis provide key insights into the standardization of future wireless network standards and design of cache-enabled base stations; which are seen as key elements for offering improved capacity, data rates and reliability in future wireless networks