Cooperative data transfers for 5G networks

The demand for higher capacity, higher data rate and larger bandwidth has driven the research and industrial world to develop next generation wireless communication technology, namely, the 5G. Among all the approaches proposed for such a high demand, only the cooperative communication approach promises to significantly improve of the performances (capacity, data rate, bandwidth, etc.) with a low cost. In this thesis, we propose a D2D communication scheme as a solution for the out-door scenario and a cooperative scheme among the access infrastructures as the in-door scenario solution. In the first part, we address the implementation of content-centric routing in a D2D architecture for Android devices based on WiFi Direct, a protocol recently standardised by the Wi-Fi Alliance. After discussing the creation of multiple D2D groups, we introduce novel paradigms featuring intra- and inter-group bidirectional communication. We then present the primitives involved in content advertising and requesting among members of the multi-group network. In addition to the communications, we also devise a mechanism to enable the devices to spontaneously establish the multi-group D2D network. Finally, we evaluate the performance of our architecture and the network formation mechanism in a real testbed consisting of Android devices. In the second part, we propose, implement and evaluate a bandwidth aggregation service for residential users that allows to improve the upload throughput of the ADSL connection by leveraging the unused bandwidth of neighboring users. The residential access gateway adopts the 802.11 radio interface to simultaneously serve the local home users and to share the broadband connectivity with neighboring access gateways. Differently from previous works, our aggregation scheme is transparent both for local users, who are not required to modify their applications or device drivers, and for neighboring users, who do not experience any meaningful performance degradation. In order to evaluate the achievable performance and tune the parameters driving the traffic balancing, we developed a fluid model which was shown experimentally to be very accurate. Our proposed scheme is amenable to efficient implementation on Linux networking stack. Indeed, we implemented it and tested in some realistic scenarios, showing an efficient exploitation of the whole available bandwidth, also for legacy cloud storage applications.