Millimeter wave (mmWave) location systems not only provide accurate positioning for location-based services, but can also help optimize network operations, for example through location-driven beam steering and access point association. In this paper, we design and evaluate localization schemes that exploit the characteristics of mmWave communication systems. We propose two range-free algorithms belonging to the broad classes of triangulation and angle difference-of-arrival. The schemes work both with multiple anchors and with as few as a single anchor, under the only assumption that the floor plan and the positions of the mmWave access points are known. Moreover, they are designed to be lightweight, so that even computationally-constrained devices can run them. We evaluate our proposed algorithms against two benchmark approaches based on fingerprinting and angles of arrival, respectively. Our results, obtained both by means of simulations and through measurements involving commercial 60-GHz mmWave devices, show that sub-meter accuracy is achieved in most of the cases, even in the presence of only a single access point. The availability of multiple access points substantially improves the localization accuracy, especially for large indoor spaces.

In this paper, we target single-anchor localization schemes for millimeter wave (MMW) systems. The schemes are designed to be lightweight, so that even computationallyconstrained devices can support them. We identify the main propagation properties of MMW signals that have an impact on localization and design three algorithms that exploit these, namely a triangulation-validation procedure, an angle differenceof-arrival approach, and a scheme based on location fingerprinting. We evaluate the algorithms by means of simulations, and draw conclusions on their robustness. We then validate our results via measurements involving commercial pre-standard 60-GHz MMW hardware. Our experiments confirm that, by relying only on a single anchor and without requiring complex signal processing at the receiver, the algorithms can localize a node with high probability, and in many cases with sub-meter accuracy. We conclude by discussing how these algorithms complement each other in terms of robustness and localization success probability.

No abstract

No abstract

No abstract