The ubiquity of portable and mobile devices has fueled the recent proliferation of wide-area wireless access technologies. Until recently, most commercial networks have offered only a small set of specialized services. These services range from one- and two-way paging, as in the commercial paging networks, to two-way electronic mail and facsimile (FAX) services. However, the tremendous growth of the World Wide Web and low-cost Internet access services has expanded the goal of wide-area wireless access providers from these specialized services to general purpose Internet connectivity.
One of the most promising wide-area wireless Internet access services is the Ricochet system from Metricom. The goal of the Ricochet system is to enable moderate speed (<100 kb/s) wireless Internet access at competitive rates. The Ricochet packet radio network achieves coverage through the deployment of a large number of inexpensive packet radios on pole tops. These pole top radios route modem packets to a wired access point (WAP), which then connects to the wired Internet. The number of hops between a modem and a WAP is variable, but is generally two or three. Ricochet employs spread spectrum, frequency-hopping technology across 160 channels in the license-free Part 15 902-928MHz ISM band, and use a Gaussian Minimum Shift Keying (GMSK) modulation scheme for the baseband frequency. The advertised maximum raw bit rate through the radios is 100 kb/s.
The purpose of this work is to provide a thorough evaluation of the Ricochet system including a rigorous performance analysis. Our results indicate a maximum TCP and UDP throughput of approximately 30 and 60 kb/s, respectively. The jitter of UDP traffic was at worst 30 ms and minimum interactive response time was 300 ms. Finally, we observed significant interference between the Ricochet and AT&T; WaveLAN devices.
The rest of the paper is organized as follows. Section 2 details the Ricochet system and provides the topology of our local deployment of the system. In Section 3 we present our experiments and performance results. Finally, in Section 4, we present some concluding remarks and observations.