Spectrum, Power, and Storage Management in Device-to-Device Networks

Device-to-device (D2D) communication allows direct data transmission among devices, which has been considered a cost-effective means to improve system performance in future networks. Potentially, D2D communication may reduce data transmission delay, increase network throughput, expand network coverage, and enhance network robustness. In spite of these promising benefits, the management of D2D wireless resources, such as spectrum, power, and storage, remains a challenging problem. In this thesis, we focus on the spectrum, power, and storage management in D2D networks. First, we focus on optimizing the transmission probabilities of D2D pairs, where the D2D pairs share spectrum resources in a random access manner. We aim at maximizing $\alpha$-fairness in D2D networks with exact topology information or with only spatial statistics. Then, we consider power control among D2D pairs in D2D networks with delayed feedback of the network state information via an on-line learning approach. We aim at maximizing the long-term averaged sum-rate of D2D pairs with performance guarantee for cellular communication. Finally, we study probabilistic caching in D2D networks in order to leverage file storage in devices. We jointly optimize spectrum allocation between cellular communication and D2D communication and file caching probabilities to maximize the logarithm utility.Ph.D