Low Complexity Node Clustering in Cloud-RAN for Service Provisioning and Resource Allocation
Low Complexity Node Clustering in Cloud-RAN for Service Provisioning and Resource Allocation
复制标题
DOI：
10.1109/glocom.2017.8254979
复制
发表时间：
2017-12
期刊：
影响因子：
--
通讯作者：
Haining Wang;Priyesh Y. Shetty;Z. Ding
中科院分区：
文献类型：
--
作者：
Haining Wang;Priyesh Y. Shetty;Z. Ding
基于拍卖的服务提供和资源分配在云-RAN无线网络体系结构和异构网络中显示出强大的潜力，以实现有效的资源共享。一个主要的技术挑战是在基于拍卖的解决方案中整合干扰约束。在这项工作中，我们将干扰约束要求转化为每个簇上的一组线性约束。我们通过开发一种新的实用的次优解决方案来解决一般的NP-Hard聚类问题，该方案能够满足我们的设计要求。我们的新算法利用了弦图的性质，并应用词典广度优先搜索(LEX-BFS)算法进行簇分裂。这种多项式时间近似算法通过在子图密度和最优聚类概率方面产生较强的性能来搜索图中的最大团，而不会受到最优解的高复杂性的影响。
Auction-based service provisioning and resource allocation have demonstrated strong potential in Cloud-RAN wireless network architecture and heterogeneous networks for effective resource sharing. One major technical challenge is the integration of interference constraints in auction-based solutions. In this work we transform the interference constraint requirement into a set of linear constraints on each cluster. We tackle the generally NP-hard clustering problem by developing a novel practical suboptimal solution that can meet our design requirement. Our novel algorithm utilizes the properties of chordal graphs and applies Lexicographic Breadth First Search (Lex-BFS) algorithm for cluster splitting. This polynomial time approximate algorithm searches for maximal cliques in a graph by generating strong performance in terms of subgraph density and probability of optimal clustering without suffering from the high complexity of the optimal solution.