MACAW

This paper presents a protocol for a specific kind of wireless network and its interactions with wired TCP. In particular, the test case is an actual network of wireless pads that are mobile and move room-to-room, with relatively small footprints of communication. The base stations are in the ceiling of the rooms, leading to the low range (since the communication is always >6 ft apart). Interestingly, the paper says that noise is not one of their major design points; I think this is somewhat unrealistic, given the noise sources present in all offices (microwaves, speakers, etc.)

The protocol designed starts off as something relatively simple: an RTS-CTS-MSG sequence.

After that, the designers use a number of scenarios to add on to their protocol, including adding ACKs, adding a Data-Send to let others know a message exchange is occurring, adding an RRTS, and so on. Although each is motivated by a specific example, the end result is a convoluted algorithm which has quite a complicated flow even in the usual case. Something so complicated gives many opportunities for flaws.

In addition, I'm not sure how handoff works in this algorithm. Since the ranges are so small, it seems like handoff would be something that occurs often.