NSF NeTS program, CNS-1318292
- Kang G. Shin (PI), University of Michigan
- Sihui Han (PhD student), University of Michigan
- Xinyu Zhang (PI), University of Wisconsin
- Xiufeng Xie (PhD student), University of Wisconsin
- URL: http://xyzhang.ece.wisc.edu/FSS
The objective of this joint project between University of Michigan and University of Wisconsin is to gain insights into coexistence of Gbps and legacy WLANs, develop an optimization framework to enable efficient spectrum sharing between them, and validate the proposed solutions in a medium-scale software-radio testbed. The emerging Gbps WLAN standard, 802.11ac, is in a draft stage but will soon be ratified and deployed along with legacy WLAN protocols. 802.11ac relies heavily on spectrum aggregation to achieve ultra-high throughput. An 802.11ac channel occupies up to 160MHz spectrum, overlapping with 8 legacy channels each with 20MHz. Such spectrum sharing between heterogeneous wireless channels, despite its importance, has not garnered enough attention from network designers. Existing network operations, such as carrier sensing, treat each 160MHz-channel as an atomic block, thus forcing the entire channel to suspend its transmission even when it is only partially occupied by, say, a 20MHz-channel. A wideband channel may thus severely underutilize its spectrum, and may even be starved when it overlaps with multiple narrowband channels with saturated traffic.
Using a testbed of COTS mobile devices, we first conduct an extensive measurement study to uncover the root cause of inefficient and unfair spectrum sharing between Gbps and legacy WLANs. Then, we develop an optimization framework which leads to decentralized spectrum sharing and channel access algorithms with provable efficiency and fairness guarantees. The algorithms will improve the MAC layer’s awareness of heterogeneous spectrum sharing, and enforce intelligent control over the PHY layer through fine-grained spectrum access and opportunistic spectrum aggregation. They will realize Gbps wireless networking even in a crowd of low-rate legacy networks/devices, and will be further generalized to whitespace networks (WiFi 2.0), where available spectrum blocks are scattered over a wide range, and hence, heterogeneous spectrum sharing becomes a central issue. Our solutions will be validated via a full-fledged implementation and experimentation on a medium-scale software radio testbed.