Cooperative Positioning based on Array Processing and Information Fusion

We are in the middle of the digital era, with more and more amazing features becoming available even in entry-level consumer devices (smartphones, tablets, wearable devices such as smart watches, etc.). This pervasive almost ubiquitous availability of interconnected devices, unconceivable until only a decade ago, is opening the doors to unprecedented applications, for which location awareness is an essential need. Unfortunately, none of the current positioning technologies alone is able to provide anywhere and anytime location capabilities, that is, to ensure service coverage in heterogeneous environments (e.g., outdoor, indoor) while offering adequate positioning accuracy. In response to such demand, this thesis investigates novel localization algorithms that can offer ubiquitous positioning capabilities. For this purpose, a novel holistic framework is proposed, based on the combined use of four dimensions of design, each focusing on a specific aspect of the whole localization task. The basic idea is to adaptively exploit all the available information under a unified multi-system localization approach. More precisely, localization capabilities can be improved by adopting, as a first ingredient, PHY-layer signal processing, which can extract valuable position-related information by exploiting different types of low-level signals. On the other hand, the positioning process can also benefit from other sources of information, namely performing some integration with additional local sensors (e.g., kinematic). Not least, higher-level data obtained through cooperation can be properly fused to improve the ultimate localization performance. This leads to a novel cross-layer architecture that allows nodes to take advantage of all the information available at different levels in each operational context, thus enabling the desirable anywhere and anytime positioning capabilities. Results obtained by simulating realistic operating scenarios show that the localization algorithms designed on the basis of the the proposed framework can achieve improved positioning performance, outperforming current state-of-the-art approaches. Furthermore, the proposed solutions exhibit reduced computational and communication complexity, thus being very attractive and easily implementable in modern mass-market devices.