Adaptive CSMA for Decentralized Scheduling of Multi-Hop Networks With End-to-End Deadline Constraints
Adaptive CSMA for Decentralized Scheduling of Multi-Hop Networks With End-to-End Deadline Constraints
复制标题
具有端到端期限约束的多跳网络的自适应 CSMA 分散调度
DOI：
--
复制
发表时间：
2021
期刊：
影响因子：
--
通讯作者：
P. Kumar
中科院分区：
文献类型：
--
作者：
Rahul Singh;P. Kumar
考虑服务多个流的多跳无线网络，其中链路之间的无线干扰约束由链路干扰图描述。流的及时吞吐量被定义为在指定期限内到达其目的地节点的该流的数据包的吞吐量，并且网络的加权及时吞吐量是具有给定的正权重集的流的加权平均值。当传输之间存在无线干扰时，这个问题尤其具有挑战性，并且通常是公开的。我们表明，具有适当尝试概率集的修改后的 CSMA 路由调度策略对于最大化加权及时吞吐量而言几乎是最佳的。该策略具有有用的属性，即单个数据包的路由调度决策仅是其位置和截止时间的函数，因此无线节点不需要了解全局网络状态。给定尝试概率，它可以很容易地由节点以分散的方式实现。进一步提供了一种基于梯度的自适应CSMA路由调度策略来确定最佳尝试概率。它沿着及时吞吐量的梯度移动并收敛到局部最大值。
Consider a multihop wireless network serving multiple flows in which wireless interference constraints between links are described by a link-interference graph. The timely-throughput of a flow is defined as the throughput of packets of that flow that reach their destination node within a specified deadline, and the weighted timely throughput of the network is their weighted average over the flows with a given set of positive weights. The problem is particularly challenging, and has generally been open, when there is wireless interference between transmissions. We show that a modified CSMA routing-scheduling policy with an appropriate set of attempt probabilities is nearly optimal for maximizing weighted timely-throughput. This policy has the useful property that the routing-scheduling decision for an individual packet is solely a function of its location and time-to-deadline, and so a wireless node does not require knowledge of the global network state. It is easily implementable in a decentralized fashion by the nodes given the attempt probabilities. A gradient-based adaptive CSMA routing-scheduling policy to determine the optimal attempt probabilities is further provided. It moves along the gradient of the timely throughput and converges to a local maximum.