Ad Hoc Routing Using Directional Antennas, Romit Roy Choudhury, Nitin Vaidya. Technical Report.
I believe this technical paper is an early version of Performance of Ad Hoc Routing using Directional Antennas, Romit Roy Choudhury, Nitin Vaidya, Journal of Ad Hoc Networks - Elsevier Publishers, November, 2004. At least the ideas and most of the content seems similar enough that I will only read the tech report, not the journal paper.
Like the previous paper I wrote about (just below), this one modifies the MAC protocol to support directional antennas, by having omni-directional listening, and directional transmitting of CTS and RTS. It also considers the impact of the directional antenna on routing, namely the DSR protocol, and how to improve DSR to support directional antennas.
The key insight is that: directional antennas improve the performance a lot when the network is random and sparse, and when the directional transmission range is significantly longer than the omni transmission range. Basically, there is a great gain when the transmission beam is narrow and focused. In other set ups, the paper does not find a great gain from directional antenna, a rather counter-intuitive result.
The paper is mostly a performance evaluation using qualnet simulation. Also, the paper takes into account the sweeping time, ie. the delay in broadcast (for DSR) incurred by transmitting the broadcast packet sequentially over the different sector (ie. sweeping) over an omni-directional broadcast to all neighbors at the same time.
An interesting idea to reduce the sweeping time is to broadcast not to all neighbors, but only to those in a few sector opposite from the one receiving sector of the broadcast. I'm curious to see whether this would not yield to route requests which miss the destination (however, it is hard to miss it, since the neighbors of the destination are as good as the destination, and since many RREQs are redundant anyway).