Beamforming towards regions of interest for multi-site mobile networks

Beamforming has been deployed in mobile phone standards starting already with 2G. The techniques have become ever more sophisticated with each iteration. 4G/LTE, for example, deploys MIMO-based beamforming. In general, direct multi-user beamforming [1], [2] – creating beams for each individual devices from mobile phone masts – is a gargantuan task. There are simply too many users, and a steady stream of accurate channel feedback [3] would be required to account people and vehicles moving around. Yet algorithms deployed to date are variations on a theme within the MIMO framework: steering the beam towards a single point [4], [5], [6]. In practise, one would like to be able to concentrate energy in a way commensurate with the locations of devices, namely target areas of interest not single points, all the while building in tolerance for movement and imprecise location information. To increase received power, LTE devices can receive signal from multiple base stations. This is a complex coordination protocol in general, and one key component to its efficiency is to have base stations target areas of importance. In this paper, we apply a technique called flexibeam, which determines beamforming weights that when applied approximate an optimal radiation pattern, so as to enable base stations to jointly concentrate energy where most devices are. The analytic framework allows tractable, numerically stable determination of beamforming weights. Aiming for areas rather than points minimises the required update rate, and reduces the communication requirement. To this end, Section II derives flexibeam from an explicit transmit beamforming perspective. We then, in Section III, illustrate its application using a Gaussian approximation to track an object in the presence of uncertainty. Afterwards, in Section III, we illustrate how energy can be concentrated around a certain area. There we take an example of devices concentrated around an area of Zurich city, and show how the target radiation pattern can be approximated by a series of Gaussian filters. We then illustrate how to determine each base station’s beamforming weights so as to concentrate energy where most devices are.