一种环境感知的无线Mesh网络自适应QoS路径选择算法

本文针对如何改善无线多跳Mesh网络的服务质量,满足无线多媒体业务对数据传输的带宽、时延、抖动的要求等问题,研究了一种基于无线信道状态和链路质量统计的MAC层最大重传次数的自适应调整算法。该算法通过对无线Mesh网络的无线信道环境的动态感知,利用分层判断法区分无线分组丢失的主要原因是无线差错还是网络拥塞导致,实时调整MAC层的最佳重传次数,降低无线网络中的分组冲突概率。基于链路状态信息的统计和最大重传策略,提出了一种启发式的基于环境感知的QoS路由优化机制HEAOR。该算法通过动态感知底层链路状态信息,利用灰色关联分析法自适应选择最优路径,在不增加系统复杂度的基础上,减少链路误判概率,提高传输效率。NS2仿真结果表明,HEAOR算法能有效减少重路由次数,降低链路失效概率,提高网络的平均吞吐率。本文提出的方法不仅能够优化MAC层的重传,而且通过发现跨层设计的优化参数实现对路径的优化选择。      

基于拥塞反馈的Semi-TCP优化方法

Semi-TCP根据无线多跳网络的特性,利用跨层设计的思想,端到端的拥塞控制调整为节点之间逐跳的拥塞控制,将传输层的拥塞控制下移到MAC层,形成递归逐跳的拥塞控制,旨在提高拥塞控制的效率和拥塞处理的即时性。现有的算法要通过增加或修改MAC控制帧的方式来实现拥塞的控制,因而需要对节点已安装的MAC层协议做适当的修改。文中提出了一种基于拥塞反馈的优化方法,不需要对传统TCP协议进行太多的修改。在EXata平台上进行了仿真,证明了该优化方法的有效性。     

一种基于Ad Hoc网络的协作MAC方案

在 IEEE 802.11b MAC层协议的研究基础上,提出了一种基于无线Ad Hoc网络的最佳中继选择策略,该策略综合考虑瞬时信道信息和节点剩余能量,能够有效防止信道条件好的节点的过度使用,并能保证系统高吞吐量。文中以饱和吞吐量和网络生存时间为性能指标,对新算法和以往协作MAC算法进行仿真和比较,结果表明新算法在网络吞吐量下降不明显的情况下,能大大增加网络生存时间。     

无线Ad hoc网络中基于节点位置的功率控制算法

为了降低无线Ad hoc网络中节点的能量消耗,该文提出了一种基于节点位置的功率控制算法(PCAP)。PCAP算法通过分析节点间的位置关系,建立节点的优化邻居集合,并对路由层报文、MAC层控制报文和其它数据类报文使用不同的功率控制策略。PCAP算法在保证网络连接性的同时能降低网络能量消耗,计算机仿真表明,PCAP算法在MAC层的吞吐量、MAC层丢包、端到端时延等方面取得较好的性能表现。     

基于交叉层设计的MIMO Ad—hoc网络的MAC层调度算法

在应用多入多出(MIMO)技术的无线Ad-hoc网络中,基于交叉层优化,设计了介质访问控制(MAC)层的优化调度算法。通过网络层、MAC层及物理层交换关键参数信息,在各无线节点分布式地进行优化调度,以选择发送节点以及在这些节点上进行流分配,选择用于发送数据的天线阵列子集,从而提高整体网络的性能。仿真结果显示,与简单单链路算法相比,交叉层设计优化算法可以显著提高MIMOAd-hoc网络的整体容量。     

基于节点接入能力的ad hoc网络按需路由协议

基于802.11协议MAC层重传策略,提出一个衡量节点接入能力的参数CAM(capacity of access to medium),以体现节点周围信道的繁忙程度及其抢占信道能力。在此基础上,联合MAC层和网络层进行跨层设计,提出了一个拥塞感知路由CAOR(congestion aware on-demand routing)协议。仿真表明:该协议能够在降低开销的前提下,显著增加网络吞吐量,并降低平均端到端的时延。     

基于MAC层速率调整的无线Mesh网络拥塞控制

为了解决无线Mesh网络的拥塞问题,从MAC层出发,设计了一套拥塞控制方案(Data Link Congestion Control,DLCC),通过自定义协议实现拥塞时的消息反馈,采用令牌桶算法限制Mesh节点下发的流量。最后分析对比了使用方案前后的网络吞吐量,验证了算法存在性能增益。     

无线局域网中的拥塞测度方法

无线局域网中,无线信道的不确定性带来了与有线网络不同的拥塞控制问题,因此传统的基于丢包的网络拥塞测度方法不能简单移植到无线网络中。文章简单介绍了无线局域网的特点,在此基础上详细分析了针对无线网络的端节点和无线接入点的拥塞测度方法,对这些方法进行了归纳和比较,最后指出在无线局域网中,MAC层拥塞测度方法是未来无线网络拥塞问题的重要研究方向。     

基于无线传感器网络的跨层拥塞控制协议

无线传感器网络(WSN)中由拥塞引起的大量分组重传以及重传多次失败后的分组丢弃会导致较长的时延、较高的分组丢失率和较多的能量消耗.为了准确探测和控制网络拥塞,提出了一种基于跨层设计的拥塞控制协议,即上行拥塞控制(UCC)协议.该协议利用节点在媒质接人控制(MAC)层中未占用的缓冲器区间大小和所预测的通信流量作为该节点的...     

基于REDP算法的动态QoS时延保障

针对无线多跳自组织网络中随机网络编码,之前提出了一种网络编码随机优先级检测调度算法(Random early detection with priority,REDP),对两条不同优先级的数据流建立Markov数学模型,推导出了引入REDP算法下数据流的单节点的时延数学表达式,该算法使得数据流能够根据自身优先级和拥塞程度调整丢包概率,从而主动丢弃数据包提前避免拥塞。文中将单节点时延问题拓展到多节点的网络拓扑结构中,提出了动态时延分配算法,该算法能够根据各个节点队列的缓存情况,动态调整各个节点的时延分配,仿真结果表明,动态时延分配算法能够实现实时数据流严格的Qo S时延保障,更利于视频、语音等实时业务流的传输。     

Performance of MIMO cross‐layer MAC protocol based on antenna selection in ad hoc networks

In this paper, we investigate the performance of a cross‐layer (physical and MAC) design for multiple‐input multiple‐output (MIMO) system that aims at maximizing the throughput of ad hoc networks by selecting the optimum antenna combination. Employing this cross‐layer design is shown to improve the overall network performance relative to the case where no antenna selection (AS) is used. To solve the node blocking problem associated with the IEEE 802.11 medium‐access control (MAC) protocol, the proposed protocol leverage the available degrees of freedom offered by the MIMO system to allow neighboring nodes to simultaneously communicate using the zero‐forcing (ZF) Bell‐labs layered space‐time (BLAST) architecture. Using the cross‐layer design, neighboring nodes share their optimum antenna selection (AS) information through control messages. Given this shared information, nodes set their decisions on the number of selected antennas based on the available spatial channels that guarantees collision‐free transmissions. At the destination node, the ZF receiver is employed to extract the desired user data while treating the data from neighboring users as interference. The performance of the proposed cross‐layer design is examined through simulations, where we show that the network throughput is significantly improved compared to conventional MAC protocols. Copyright © 2010 John Wiley & Sons, Ltd.     

一种基于消息缓存的多跳协作机制*

提出一种基于消息缓存的多跳协作机制(ICC机制),该机制不需要依靠准确的网络状态信息和复杂的协作节点选择算法,而是通过利用RTS/CTS交互消息获得相隔两跳的节点MAC地址,并在数据帧中添加下两跳的节点MAC地址,使得相隔两跳的节点可靠的接收到节点发送的信号,并对接收到的信号进行合并,从而获得分集为2的增益。     

一种面向实时业务的Ad Hoc网络MAC协议

为降低AdHoe网络中实时业务的端到端时延,提出了基于802．11DCF的改进协议。协议采用3种机制降低实时业务的时延：面向路径的连续转发机制将RTS中的转发信息携带在ACK中发送,给实时业务提供较高的接入优先级;标签交换机制使得中间节点可以在MAC层获取转发信息,加快了实时业务数据包的转发速度;重传控制机制减少了无效传输的超时数据包。仿真结果表明,在重负载条件下,改进协议中实时业务的时延比802．11DCF有大幅度的下降,网络吞吐量也有所提高。     

基于名声、可用带宽和最小跳数的多量度自适应Ad hoc路由协议

通过对实际Ad hoc网络中节点通信情况的研究,考虑到网络中节点的自私性以及单纯使用最小跳数选路方法的缺陷,综合考虑名声、可用带宽和最小跳数,该文提出了一种新的参数量度。基于这个量度,借鉴TOPP(Trains Of Packet Pairs)测量可用带宽的思想,设计了一种快速测量逐跳可用带宽的方法,并通过MAC层和网络层的跨层设计,提出了一个基于以上混合量度的路由协议(Ad hoc Routing Based on Fame,Available Bandwidth and Minimum Hops, ARBFAM)。通过使用该协议在不同场景的网络环境中仿真证明,该路由协议在节点公平性,网络的拥塞控制和负载均衡,网络的平均寿命,网络的端到端吞吐量,平均端到端延迟方面较DSR都有较大的改善。     

A cross-layer framework for position-based routing and medium access control in heterogeneous mobile ad hoc networks

Many Mobile Ad Hoc Networks (MANETs) are composed of nodes with heterogeneous communication capabilities in terms of transmission power, data rate, battery life, reliability, etc. Heterogeneous MANETS (H-MANETs) often encounter problems in both Medium Access Control (MAC) and routing layers due to link asymmetry. But meanwhile, H-MANETs have many advantages including: scalability, robustness, and connectivity. In this paper we present a two-layer approach which spans the MAC and network layers for H-MANETs to minimize the problems caused by link asymmetry and exploit the advantages of H-MANETs simultaneously. At the network layer, we design a hierarchical Location service (LS) protocol and a node density adaptive position-based routing protocol. At the MAC layer we propose a new MAC protocol which uses two communication channels and closely ties in with the network layer to solve the MAC problems caused by link asymmetry. We evaluate our approach using both probabilistic analysis and simulations.     

Signal processing in random access

In this paper, a cross-layer view for roles of signal processing in random access network and vice versa is presented. The two cases where cross-layer design has a quantifiable impact on system performance are discussed. The first case is a small network (such as wireless LAN) where a few nodes with bursty arrivals communicate with an access point. The design objective is to achieve the highest throughput among users with variable rate and delay constraints. The impact of PHY layer design on MAC protocol is examined and illustrates a tradeoff between allocating resources to the PHY layer and to MAC layer. The second case, in contrast, deals with large-scale sensor networks where each node carries little information but is severely constrained by its computation and communication complexity and most importantly, battery power. This paper emphasizes that the design of signal processing algorithms must take into account the role of MAC and the nature of random arrivals and bursty transmissions.     

基于SMAC的无线传感器网络MAC协议的分析与优化

首先对MAC协议进行了相关介绍。然后重点介绍了一种基于竞争的无线传感器网络MAC层协议S-MAC协议。其核心是提出了一种新的无线传感器网络的MAC协议设计方案。基于动态调整占空比的思想,提出了ATC-SMAC协议。该协议在S-MAC协议的基础上改进了固定占空比的劣势,根据每个节点上的数据包的平均延迟调整占空比。通过动态地调整每个节点的占空比,使不同流量的节点拥有不同的工作时间,协议根据不同节点的流量情况自适应地对其占空比进行调整。经过仿真试验,得到ATC-MAC在网络端对端延迟、能量消耗以及吞吐量方面较S-MAC协议都有比较明显的提高。     

Routing in the frequency domain

The design of single transceiver based multi-channel multi-hop wireless mesh networks focuses on the trade-off between rapid neighbor synchronization and maximizing the usage of all available channels. Existing designs are confined to the MAC layer and scale poorly as the network grows in coverage and density. We recently proposed Dominion as a cross-layer architecture that includes both medium access control and routing. Dominion eliminates the need for neighbor synchronization at the MAC layer and pushes the intelligence up the network stack. At the MAC layer, a node switches channels according to a deterministic schedule which guarantees that a node converges with each of its neighbors periodically. At the network layer, the channel-hopping aware routing substrate routes traffic along the frequency domain, i.e., packets along a multi-hop route generally traverse via multiple channels. In this paper, we present the complete design, analysis and evaluation of Dominion and make four new contributions. Firstly, we extend Dominion to support goal-oriented routing: source nodes can locally choose to maximize throughput or minimize end-to-end latency without requiring any changes in the network. Secondly, we describe a technique that eliminates intra-flow interference. In absence of extrinsic interference, Dominion now allows network flows to maintain constant throughput and deterministic end-to-end latencies irrespective of distance. Thirdly, via theoretical modeling and analysis, we provide expected throughput and end-to-end latencies for network flows. Finally, via extensive QualNet simulations we show that Dominion achieves 1064% higher throughput than IEEE 802.11 while being 299% fairer.