[论文解读] Joint Access and Backhaul Resource Management in Satellite-Drone Networks: A Competitive Market Approach

Ye Hu, Mingzhe Chen+1 more|arXiv (Cornell University)|Aug 29, 2019

Satellite Communication Systems参考文献 35被引用 6

一句话总结

本文提出了一种基于竞争市场的框架，用于在集成卫星-无人机网络（ISDNs）中联合管理接入与回传资源，将用户、基站和基础设施建模为买家与卖家，以实现瓦尔拉斯一般均衡。通过使用重型球算法，确保收敛至最优资源分配，使高速率接入链路数量提升两倍，高速回传链路容量提升三倍。

ABSTRACT

In this paper, the problem of user association and resource allocation is studied for an integrated satellite-drone network (ISDN). In the considered model, drone base stations (DBSs) provide downlink connectivity, supplementally, to ground users whose demand cannot be satisfied by terrestrial small cell base stations (SBSs). Meanwhile, a satellite system and a set of terrestrial macrocell base stations (MBSs) are used to provide resources for backhaul connectivity for both DBSs and SBSs. For this scenario, one must jointly consider resource management over satellite-DBS/SBS backhaul links, MBS-DBS/SBS terrestrial backhaul links, and DBS/SBS-user radio access links as well as user association with DBSs and SBSs. This joint user association and resource allocation problem is modeled using a competitive market setting in which the transmission data is considered as a good that is being exchanged between users, DBSs, and SBSs that act as "buyers", and DBSs, SBSs, MBSs, and the satellite that act as "sellers". In this market, the quality-of-service (QoS) is used to capture the quality of the data transmission (defined as good), while the energy consumption the buyers use for data transmission is the cost of exchanging a good. According to the quality of goods, sellers in the market propose quotations to the buyers to sell their goods, while the buyers purchase the goods based on the quotation. The buyers profit from the difference between the earned QoS and the charged price, while the sellers profit from the difference between earned price and the energy spent for data transmission. The buyers and sellers in the market seek to reach a Walrasian equilibrium, at which all the goods are sold, and each of the devices' profit is maximized. A heavy ball based iterative algorithm is proposed to compute the Walrasian equilibrium of the formulated market.

研究动机与目标

• 解决在集成卫星-无人机网络（ISDNs）中，卫星、地面与用户无线接入链路的联合资源分配问题。
• 将用户关联与资源分配建模为一个竞争市场，其中数据传输为商品，服务质量（QoS）为质量，能耗为交易成本。
• 实现瓦尔拉斯一般均衡，即所有商品均被售出，所有参与者均实现利润最大化，从而确保系统整体效率。
• 设计一种基于重型球的迭代算法，保证收敛至均衡状态，且收敛速度优于次梯度方法。
• 在真实ISDN部署场景下，评估高速率接入链路与回传链路的性能增益。

提出的方法

• 将ISDN资源分配问题形式化为一个竞争市场，其中用户和基站作为买家，卫星、MBSs和SBSs作为卖家。
• 将数据传输定义为商品，QoS作为其质量，能耗作为交易成本。
• 将买家的利润定义为获得的QoS与支付价格之间的差值，将卖家的利润定义为收到价格与能耗成本之间的差值。
• 应用具有明确定义步长的重型球算法，迭代更新价格与资源分配，以逼近瓦尔拉斯一般均衡。
• 依赖凸松弛与强对偶性，确保原始问题最优解的收敛性。
• 采用时隙模型，结合动态卫星-基站距离与路径损耗模型，计算时变数据速率。

实验结果

研究问题

• RQ1 在具有多层基础设施的集成卫星-无人机网络中，如何实现接入与回传资源的最优联合管理？
• RQ2 竞争市场框架能否有效建模ISDN中用户、基站与回传提供商之间的交互关系？
• RQ3 何种算法方法可确保在此市场模型中收敛至瓦尔拉斯一般均衡？
• RQ4 与传统次梯度方法相比，所提方法在收敛速度与性能方面有何差异？
• RQ5 在所提框架下，高速率接入与回传链路的性能增益可达多少？

主要发现

• 所提出的重型球算法使数据速率超过40 Mbps的无线接入链路数量实现两倍增长。
• 该算法使数据速率高于1.6 Gbps的回传链路数量实现三倍增长。
• 与次梯度方法相比，该算法收敛速度提升10%，且在明确定义的步长下保证收敛。
• 分析结果证实，该算法收敛至瓦尔拉斯一般均衡，确保市场出清与所有参与者的利润最大化。
• 对整数非线性规划进行凸松弛，确保存在整数最优解，从而验证了方法的有效性。
• 在Slater条件成立下，强对偶性成立，证实对偶问题的解与原始最优值一致。