An Offense Against Wireless Carrier Sense by: Daniel Burgener Komal Pal
Why a theoretical model? Problems with carrier sense shown experimentally: V Bharghavan et. al. “MACAW: A Media Access Protocol for Wireless LANs” Y. Cheng “Jigsaw: Solving the Puzzle of Enterprise 802.11 Analysis” K. Jamieson et. al. “Understanding the Real-World Performance of Carrier Sense” M Vutukuru et. al. “Harnessing Exposed Terminals in Wireless Networks
Poor Assumptions Only two contending senders “The paper's conclusions would likely not hold for large numbers of nearby senders” (section 2) Assumes the senders always agree on choice between concurrency and mulitplexing Most of the problems with Carrier Sense arise under disagreements
Other models MACAW ZigZag All analysis compares performance to pure multiplexing and pure concurrency. Of course carrier sense is better.
Some Questions for the Defense From the paper: “carrier sense provides nearly optimal throughput in the common case” What is the common case? From the paper: “There are also no dramatic losses of efficiency due to differing channel conditions among the nodes, the differences that Karn first warned of” Can you explain and justify this?
The Experiment Only did one simple, small scale experiment The point of the experiment: To put the word “experiment” in the paper The experimental results agree with their model for the one situation tested What about other situations? Is this that mythical “common case”? The short-range long-range distinction is based on packet loss, not actual distance
Defense Conclusions Carrier sense does work, in a large, important class of networks See paper for discussion of other issues like threshold robustness Room for improvement in corner cases, but not much room for overall performance A fresh look at modeling can help us balance out the idiosyncrasies in experimental wireless work
The actual conclusions of the paper Carrier Sense works well in the best-case scenarios when it wouldn't matter anyways We really like ABR! Let's just stop doing research now, everything is good enough