Cooperative Spectrum Sharing: A Contract-Based Approach
Cooperative Spectrum Sharing: A Contract-Based Approach
复制标题
DOI：
10.1109/tmc.2012.231
复制
发表时间：
2014
影响因子：
7.9
通讯作者：
Lingjie Duan;Lin Gao;Jianwei Huang
中科院分区：
文献类型：
--
作者：
Lingjie Duan;Lin Gao;Jianwei Huang
为所有参与方提供经济激励对于动态频谱接入的成功至关重要。协作频谱共享是实现这一点的一种有效方式，其中次要用户（SU）中继主要用户（PU）的业务，以换取用于SU自己的通信的专用频谱接入时间。本文研究了不完全信息下的协作频谱共享问题，其中SU的无线特性是私有信息，不为PU所知。我们使用契约理论将PU-SU互动模型化为劳动力市场。在契约理论中，雇主一般在雇佣前并不完全了解雇员的私人信息，需要在不完全信息下向雇员提供契约。在我们的问题中，PU和SU分别是雇主和雇员，并且合同由表示频谱接入时间的组合的一组项目组成（即，奖励）和中继功率（即，贡献）。我们研究了弱不完全信息和强不完全信息情形下的最优契约设计。在弱不完全信息的情况下，我们表明，PU将最佳地雇用最有效的SU和PU达到相同的最大效用，在完整的信息基准。然而，在强不完全信息的情况下，PU也可能保守地雇用效率较低的SU。我们进一步提出了一个分解和比较（DC）的近似算法，实现了接近最优的合同。我们进一步表明，PU的平均效用损失由于次优DC算法和强烈的不完全信息是相对较小的（小于2%和1.3%，分别在我们的数值结果与两个SU类型）。
Providing economic incentives to all parties involved is essential for the success of dynamic spectrum access. Cooperative spectrum sharing is one effective way to achieve this, where secondary users (SUs) relay traffics for primary users (PUs) in exchange for dedicated spectrum access time for SUs' own communications. In this paper, we study the cooperative spectrum sharing under incomplete information, where SUs' wireless characteristics are private information and not known by a PU. We model the PU-SU interaction as a labor market using contract theory. In contract theory, the employer generally does not completely know employees' private information before the employment and needs to offers employees a contract under incomplete information. In our problem, the PU and SUs are, respectively, the employer and employees, and the contract consists of a set of items representing combinations of spectrum accessing time (i.e., reward) and relaying power (i.e., contribution). We study the optimal contract design for both weakly and strongly incomplete information scenarios. In the weakly incomplete information scenario, we show that the PU will optimally hire the most efficient SUs and the PU achieves the same maximum utility as in the complete information benchmark. In the strongly incomplete information scenario, however, the PU may conservatively hire less efficient SUs as well. We further propose a decompose-and-compare (DC) approximate algorithm that achieves a close-to-optimal contract. We further show that the PU's average utility loss due to the suboptimal DC algorithm and the strongly incomplete information are relatively small (less than 2 and 1.3 percent, respectively, in our numerical results with two SU types).