IMDEA Networks Institute Publications Repository

Mitigating Lateral Interference: Adaptive Beam Switching for Robust Millimeter-Wave Networks

Steinmetzer, Daniel and Loch, Adrian and García-García, Amanda and Widmer, Joerg and Hollick, Matthias (2017) Mitigating Lateral Interference: Adaptive Beam Switching for Robust Millimeter-Wave Networks. In: The 1st ACM Workshop on Millimeter-Wave Networks and Sensing Systems (mmNets 2017), 16 October 2017, Snowbird, Utah, USA.

[img] PDF (Mitigating Lateral Interference: Adaptive Beam Switching for Robust Millimeter-Wave Networks) - Published Version
Download (1326Kb)


Putting into practice "pseudo-wire" links in wireless millimeter-wave (mm-wave) networks is challenging due to the significant side lobes of consumer-grade phased antenna arrays. Nodes should steer their beams such that they maximize the signal gain but also minimize interference from lateral directions via both their main lobe and their side lobes. Most importantly, interference can be caused by parallel operation of incompatible standards such as WiGig and IEEE 802.11ad and may change very fast. This timing requirement, prevents the use of existing beam switching solutions to mitigate interference. In this paper, we present an adaptive beam switching (ABS) mechanism that can deal with the above timescale issue in rapidly changing interference scenarios. Instead of performing a full beam sweep, the key idea is to only probe beampatterns at the receiver which are likely to avoid interference. In contrast to earlier work, our mechanism does not require any location information nor a detailed shape of the beampatterns. We exploit similarities among side lobes of beampatterns to estimate the performance of all beampatterns without sending extensive probes. To evaluate our mechanism in practice, we develop a customized research platform that allows us to control the beam-selection on low-cost IEEE 802.11ad routers. Experimental results with WiGig transceivers as interference source show that our adaptive beam switching mechanism achieves an average throughput gain of 60% and decreases the training time by 82.4% compared to the original IEEE 802.11ad behavior.

Item Type: Conference or Workshop Papers (Paper)
Depositing User: Adrian Loch
Date Deposited: 05 Sep 2017 09:32
Last Modified: 27 Oct 2017 07:39

Actions (login required)

View Item View Item