Friday, March 16, 2007

Medium Access Control Protocols Using Directional Antennas in Ad Hoc Networks

Medium Access Control Protocols Using Directional Antennas in Ad Hoc Networks, Young-Bae Ko, Vinaychandra Shankarkumar, and Nitin Vaidya, in Proc. of IEEE Infocom 2000.

The second reviewer had me wondering if I usually write the full name or only the first initial. If you read the paper, see how his (her?) email handle is much more convenient.

The paper is a look into how to modify the 802.11 protocol to support directional antenna. Basically, due to deafness and blind terminal issues, you still need to send omni-directional control packets, but in some situation, you should send some that are directional, using only on antenna sector to transmit. First: you should always use directional antenna for the transmission. The paper details two schemes to use directional control messages.

The paper is relatively simple in its insights, but that's what you get for blazing a new trail, you have to get the low hanging fruits first. The directional MACs proposed in the paper (reasonably) assume that the location (in terms of which sector is best to use to transmit/receive to them) of the neighbors is known.

The two schemes are: 1-using directional RTS (so that you don't block nodes away from where you are transmitting); 2-using both ORTS and DRTS depending on the current status of the corresponding sector.

The performance evaluation is rather basic, using a 5x5 grid with some simple scenarios. I would have enjoyed a more realistic set-up. For instance, directional antenna performance is different on a grid, because nodes line up within a sector, so it artificially increase the interference. Also, I don't know if the baseline comparison is directional for 802.11 or not. It looks like what is being compared is:
-basic 802.11 with omni-directional antenna; and -proposed MAC with directional antenna. That is 2 parameters here (basic 802.11 vs proposed MAC, and omni vs. directional) in the simulation, while the results are used to quantify and discuss the impact of only the second parameter.

The interference level in the network for bare 802.11 would significantly decrease when using directional antenna without modifying the MAC, thus improving the performance. I don't see this explicit here. It would have been nice to quantify the impact of pure directional antennas, as there is an area of space where the interference of an on-going connection is strong enough to disturb another connection attempt nearby, but not strong enough for the nodes to successfully decode the MAC packets and update the NAV tables. For those, using a basic 802.11 with directional antennas and appropriate power control would go a long way improving performance.

No comments: