Smartphones have become an essential part of our society. The benefits of having an always present, highly capable device cannot be overstated. As more aspects of our life depend on our smartphones, it is more important than ever to ensure the availability of those devices. However, their big advantages also come with big risks. The fact that we have our smartphones with us all the time means that it is easier than ever to collect our information, sometimes without our consent. In this dissertation, we study the two pressing concerns in cellular communications: energy efficiency and privacy protection. We focus on LTE networks, the current most advanced global standard for cellular communications.

In the first part of the dissertation, we study the energy efficiency problem from both device and network perspectives. From the device point of view, we introduce a new angle to address the battery life concern. We recognize that the value of battery for the users is not always the same, and that it depends on the user usage. We also identify, and show in real network, diversity of usage, the phenomenon that at any instant, there is a diverse distribution of smartphone usage among cellular users. We propose ``Battery Deposit Service'' (BDS), a cooperative system which makes use of device-to-device (D2D) communications underlaying cellular networks to provide energy sharing in the form of load sharing. We design BDS to take advantage of diversity of usage to maximize the utility of smartphone battery. We show that our system increases battery life of cellular users, at almost no cost to the rest of the network. BDS is designed to be compatible to LTE architecture.

From the network point of view, we design an energy efficient D2D relay system underlaying LTE networks. We minimize transmission power of smartphones by considering relay selection, resource allocation and power control. The overall problem is prohibited due to its exponential search space. We develop a divide-and-conquer strategy which splits the overall problem into small sub-problems. We relate these sub-problems to well-studied graph theoretic problems, and take advantage of existing fast algorithms. We show that our algorithms meet the runtime requirement of real-time LTE operations.

In the second part of the dissertation, we address a privacy concern in LTE networks. In particular, we show that user location can be leaked in current LTE paging architecture. We propose a mechanism based on signal processing to remedy this vulnerability. Our method makes use of physical layer identification, which are low-power tags embedded on the wireless waveform, to signal paging messages to user devices. We show that our method is stealthy and robust, and that it mitigates the aforementioned privacy issue.