Multi-bid spectrum auctions in dynamic spectrum access networks with spatial reuse

In this paper, we propose an auction based spectrum allocation scheme considering a primary user (PU), who is willing to sell the unused spectrum bands to multiple secondary users (SUs). Each SU is allowed to place multiple bids-each bid is a tuple consisting of amount of bandwidth requested and the corresponding price the SU is willing to pay for that bandwidth. The bids from a SU follow a typical demand curve which we model using the Sigmoid function. Once all bids are placed, the PU determines the winner set such that: i) mutually interfering SUs are not allocated the same channel and ii) a channel can be sold multiple times as long as the corresponding SUs do not interfere with each other. Instead of using the `unit price' which is a commonly used heuristic for ordering the bids from a bidder, we use an exponentially scaled version of the unit price which we term as the normalized bid price. The average normalized bid price is used for ordering the bidders. The proposed auction scheme takes O(nlogn) time while improving the spectrum reuse and generating close to maximum revenue. We validate the proposed allocation process through simulation experiments and show what fraction of secondary users get the bandwidth and how many times the bandwidth is reused. We also show that the proposed heuristic abides by the economic properties of auctions.