Physical Concerns for Cross-Layer Prototyping and Wireless Network Experimentation


Authors:

Ketan Mandke, Robert C. Daniels, Soon-Hyeok Choi, Scott M. Nettles, and Robert W. Heath, Jr.

Reference:

Proceedings of the Second ACM International Workshop on Wireless Network Testbeds, Experimental Evaluation and Characterization, Montreal, Canada, pp. 11-18, September 2007

Abstract:

The performance of a wireless network protocol is inseperably linked to the physical layer algorithms on which it is built, the hardware used to implement the radio, and the wireless environment in which it operates. This paper identifies three features of wireless networking protocols impacted by these lower-level characteristics which are often overlooked or misunderstood by many researchers developing wireless protocols or using testbed-based evaluation methods. These features are temporal scaling, measurement reciprocity, and cross-layer adaptation. Temporal scaling refers to the time resolution with which events, such as broadcast or feedback, occur in the wireless network. This feature is tightly coupled with processing time at the physical layer and time selectivity in the wireless channel. Measurement reciprocity is an assumption used to estimate parameters of the forward link of a bidirectional communication channel, based on observations from the reverse link. This assumption directly depends on the interference properties and hardware symmetry of nodes in a wireless network. System adaptation, based on reciprocity or feedback, inevitably requires careful scrutiny of power and rate control applied to physical wireless devices. This paper also provides recommendations to guide researchers in setting up interesting and useful wireless experiments. Three concerns for wireless experimenentation are addressed, namely: ambient interference, RF hardware profiling, and fading properties of the wireless channel. The motivation for this paper stems from experience prototyping and experimenting with Hydra, a wireless cross-layer testbed developed at the University of Texas at Austin.