Scalable Multicast in Highly-Directional 60 GHz WLANs

Abstract

60 GHz bands target multi-gigabit rate applications such as high definition video streaming. Unfortunately, to provide multicast service, the strong directionality required at 60 GHz precludes serving all clients in a multicast group with a single transmission. Instead, a multicast transmission is comprised of a sequence of beam-formed transmissions (a beam group) that together cover all multicast group members. In this paper, we design, implement, and experimentally evaluate Scalable Directional Multicast (SDM) as a technique to (i) train the access point with per-beam per client RSSI measurements via partially traversing a codebook tree. The training balances the objectives of limiting overhead with collecting sufficient data to form efficient beam groups. (ii) Using the available training information, we design a scalable beam grouping algorithm that approximates the minimum multicast group data transmission time. We implement the key components of SDM and evaluate with a combination of over the-air experiments and trace-driven simulations. Our results show that the gains provided by SDM increase with group size and provide near-optimal group selection with significantly reduced training time, yielding up to 1.8x throughput gains over exhaustive-search training and grouping.