ABSTRACT
Contents 3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .62
3.1.1 Related Work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .63 3.1.2 Contributions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .64 3.1.3 Organization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .64
3.2 A Framework for Opportunistic Spectrum Sharing . . . . . . . . . . . . . . . . . .65 3.2.1 Effective Nonopportunistic Bandwidth . . . . . . . . . . . . . . . . . . . . . . .67
3.3 Problem Formulation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .68 3.3.1 Color Decoupling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .70
3.4 Proposed Algorithms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .71 3.4.1 Optimal Solutions: Benchmark . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .71 3.4.2 Distributed Greedy Algorithm . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .72 3.4.3 Distributed Fair Algorithm . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .73 3.4.4 Randomized Distributed Algorithm . . . . . . . . . . . . . . . . . . . . . . . . . . .74
3.5 Numerical Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .75 3.5.1 Simulation Setup . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .76 3.5.2 Performance under the Snapshot of the Network . . . . . . . . . . . .78 3.5.3 Performance under Time-Varying Channel Availability . . . . . .82
3.6 Conclusions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .84 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .85
3.1 Introduction The radio spectrum is among the most heavily regulated and expensive natural resources around the world. In Europe, the 3G spectrum auction yielded 35 billion dollars in England and 46 billion in Germany. The question is whether spectrum is really this scarce. Although almost all spectrum suitable for wireless communications has been allocated, preliminary studies and general observations indicate that much of the radio spectrum is not in use for a significant amount of time, and at a large number of locations. For instance, experiments conducted by Shared Spectrum Company indicate 62% percent “white space” (unused space) below the 3GHz band, even in the most crowded area near downtown Washington, DC, where both governmental and commercial spectrum usage are intensive.[1] In the experiment, a band is counted as white space if it is wider than 1MHz and remains unoccupied for 10 minutes or longer. Furthermore, spectrum usage levels vary dramatically in time, geographic locations, and frequency. A lot of the precious spectrum (below 5GHz), that is worth billions of dollars, and is perfect for wireless communications sits there silently. The large proportion of white space indicates that opportunistic or dynamic spectrum usage may significantly mitigate the spectrum scarcity.
