Multi-Channel MAC for Ad Hoc Networks: Handling Multi-Channel Hidden terminals Using A Single Transceiver. J. So, N. Vaidya, in Proc. of ACM MobiHoc 2004, Roppongi, Japan.
I read the paper, but it was a short read: it has been cited and the principles explained in a few other articles, so I just basically had to check the paper existed. But the read was fast: I had an idea what it was about, I did not have to be convinced and scratch my head too much. Plus, the idea is simple and elegant, as more often than not with Nitin.
What is not usually explained: the actual channel selection (how two nodes decide during the ATIM window to pick one channel over the other; I kinda assumed it would be a round-robin or random or something uniform like that. It is actually based on the number of previous nodes requesting the channel for transmission and a list of preference based on channel quality). And the performance analysis: mostly a comparison with DCA, a multi-channel MAC with multiple interfaces (MMAC uses only one interface). Good stuff all around.
The unrelated note: Roppongi means 6 trees.