异构无线多媒体传感器网络分组调度算法研究

发布时间：2018-06-18 02:27

本文选题：异构无线多媒体传感器网络 + 分组调度　；参考：《华南理工大学》2014年博士论文

【摘要】：无线多媒体传感器网络（WMSN）是一种新型的无线传感器网络，是在传统无线传感器网络的基础上引入了图像、音频、视频等多媒体信息感知功能。异构的无线多媒体传感器网络中包含标量节点、低端图像节点和高端视频节点等多种类型的节点，具有更好的性能，在可扩展性、覆盖范围和可靠性方面表现突出。异构无线多媒体传感器网络中的多种节点会产生各种类型的数据，这些数据的紧急程度各不相同，每一种数据的分组都会根据紧急程度设置不同的有效生存期。各种类型的数据对网络的服务质量要求之一就是：在有效生存期之内将分组运送到达目标节点。WMSN要满足这一要求面临了两个挑战：第一，因为传感器节点不能太大，使得它无法拥有较强的运算能力，，提供较大的内存和带宽，以及长时间的供电；第二，所有分组的目标节点都是汇聚节点，紧急程度相同的分组其有效期设置也应相同。而距汇聚节点远的节点产生的分组较之距汇聚节点近的节点产生的分组需要经过更多的中间节点才能到达汇聚节点。这样，远端节点的分组在中间节点停留的时间必须小于近端节点的分组。分组调度是解决服务质量要求问题的一个重要的手段。通常为不同服务质量要求的分组分配不同的优先级，再根据优先级来排队转发分组。一方面，如果一个分组在剩余有效期内不能到达汇聚节点，将会被丢弃，将导致较高的丢包率。而在WMSN中，传感器节点资源有限，所以快到期的分组，需要分配较高的优先级。另一方面远端节点的分组在中间节点停留的时间必须小于近端节点的分组，所以由距汇聚节点远的节点产生的分组应分配更高的优先级。根据以上的考虑WMSN要求分组调度算法要根据每一个分组剩余有效期的长短和其源节点距离汇聚节点的远近来分配优先级。目前提出的各类调度算法基本上都没有满足这一要求。所以本文提出了一个新的分组调度算法：无线区分队列服务（WDQS）。WDQS根据每一分组必须离开本节点的最后时间（LDT）对分组进行排队转发。LDT由分组的有效生存期（这是可以人为设置的）减去分组到达本节点已经经过的时延（这是已知量）和离开本节点后到达汇聚节点还须经过的时延（这是未知量）来决定。分组的有效期减去已经经过的时延就是分组的剩余有效期；已经经过的时延和还须经过的时延之和对应着分组的源节点与汇聚节点之间的距离（我们的研究是建立在地理位置路由之上的）。所以WDQS是根据分组的剩余有效期和其源节点距离汇聚节点的远近来分配优先级。如何设置分组的有效生存期和如何计算未知的时延是较难解决的问题。本文具体的贡献如下：第一，提出了根据LDT排队转发分组的方案。这一方案是根据分组的剩余有效期和其源节点与汇聚节点之间的距离来分配分组优先级的。第二，利用WMSN的特性，找到了一种对分组的剩余未知时延的估计方法。这一方法不会增加额外的费用和能耗。第三，研究了分组有效生存期的设置方案，分别得出了实时分组和标量分组的有效生存期的设置条件。本文从研究单数据源理想情况下的分组有效期设置方案入手，在其基础上提出了多数据源理想情况下的分组有效期设置方案，并在前两种理想情况研究的基础上提出了任意数据源的通用分组有效期设置条件表达式。分别在满足和不满足条件表达式时对网络性能进行了详细的分析，找到了在不满足条件表达式时，实时分组开始产生丢包的时间点。研究从简单到复杂，逐层深入，对各种情况都进行了充分的讨论。第四，WMSN中的应用分为两种类型：周期性数据采集型和突发性事件驱动型。CBR过程和poisson过程比较适用于描述周期性的数据，而FBM过程适用于描述突发性数据。本文把CBR，poisson和FBM过程分别作为分组的到达过程代入所得到的通用分组有效期设置条件表达式，得出了这三种情况下具体的分组有效期设置条件表达式。并对这三个表达式进行了充分的仿真实验。仿真实验的结果验证了所推导出的这三个分组有效期设置条件表达式合理、有效，这说明了通用分组有效期设置条件表达式合理，正确，适用于各种类型的无线多媒体传感器网络应用。第五，WDQS算法的分组有效期配置和具体的网络状况相关，人工配置不易实现。异构无线多媒体传感器网络中，逻辑上可以分为三种类型的节点：汇聚节点，源节点和中间节点。本文提出了WDQS算法在三种节点上的具体实现，得出了获取算法各参数的方案，实现了参数的自动配置。为算法从理论方案到具体实现提供了可能。
[Abstract]:Wireless multimedia sensor network (WMSN) is a new kind of wireless sensor network. Based on the traditional wireless sensor network, it introduces multimedia information perception functions, such as image, audio, video and so on. Heterogeneous wireless multimedia sensor networks include scalar nodes, low-end image nodes and high-end video nodes. Nodes have better performance and are outstanding in scalability, coverage and reliability.
A variety of nodes in the heterogeneous wireless multimedia sensor network produce various types of data. The urgency of these data is different. Each data group will set up different effective lifetime according to the emergency. One of the quality requirements of various types of data to the network is that within the effective lifetime The packet delivery arrival target node.WMSN is faced with two challenges: first, because the sensor nodes can not be too large to have strong computing power, provide larger memory and bandwidth, and long time power supply; second, all the target nodes of all groups are converging nodes, the emergency of the same score is the same. The time setting of the group should be the same. The packet generated by the node far away from the convergence node needs to pass through more intermediate nodes to reach the convergence node. So the time of the remote node's grouping in the middle node must be smaller than that of the near end node.
Packet scheduling is an important means to solve the problem of quality of service. It usually assigns different priority to the groups of different service quality requirements, and then queues the packets according to the priority. On the one hand, if a packet can not reach the aggregation node in the remaining validity period, it will be discarded and will lead to a higher packet loss rate. In WMSN, the node resources of the sensor nodes are limited, so the fast maturing packets need to be assigned a higher priority. On the other hand, the time for the remote node to stay in the middle node must be less than the packet in the near end node, so the packet generated by the remote node from the remote node should be assigned a higher priority. According to the above consideration, the WMSN should be considered. The packet scheduling algorithm should allocate priority according to the length of the validity period of each packet and the distance between the source node and the sink node.
In this paper, a new packet scheduling algorithm, wireless differentiated Queuing service (WDQS).WDQS, is proposed in this paper, which is based on the last time (LDT) of each packet which must leave the node. The packet is queued and forwarded to the effective lifetime of the.LDT (this is human set. Subtracting the time delay (this is the known amount) and the delay (this is the unknown amount) that the node has passed after the node has passed (this is the known amount). The time delay that has been passed is the remaining validity period of the packet, and the time delay passed and the time delay that has to pass through the packet is corresponding to the time delay. The distance between the source node of the packet and the aggregation node (our research is based on the geographical location routing). So WDQS is based on the residual validity of the group and the distance from the distance of the node distance to the node. How to set up the effective lifetime of the group and how to calculate the unknown delay is a difficult problem to solve. The specific contributions of this article are as follows:
First, a packet forwarding scheme based on LDT queuing is proposed. This scheme is based on the residual validity of the packet and the distance between the source node and the aggregation node to assign the packet priority.
Second, using the characteristics of WMSN, we find a method for estimating the remaining unknown time delay of packets. This method does not increase additional cost and energy consumption.
Third, the setting scheme of effective subsistence period is studied, and the setting conditions for the effective lifetime of real time grouping and scalar grouping are obtained. On the basis of the two ideal case studies, a conditional expression for the setting of a general packet validity period for arbitrary data sources is proposed. The network performance is analyzed in detail when the conditional expression is satisfied and the conditional expression is not satisfied. The time points of the packet loss are found when the conditional expression is not satisfied. Layer by layer, we have discussed all kinds of situations thoroughly.
Fourth, the application in WMSN is divided into two types: periodic data acquisition and sudden event driven.CBR processes and Poisson processes are suitable for describing periodic data, while FBM processes are suitable for describing burst data. This paper takes the CBR, Poisson, and FBM processes as the arrival process of the group into the general groups obtained. The term expression of the validity period is set, and the conditional expression of the set term setting of the three cases is obtained. The full simulation experiments are carried out on the three expressions. The result of the simulation experiment verifies the rational and effective expression of the set condition of the three groupings. This shows that the validity period of the general group is set up. The conditional expression is reasonable and correct, and suitable for all kinds of wireless multimedia sensor network applications.
Fifth, the effective allocation of the WDQS algorithm is related to the specific network condition, and the artificial configuration is not easy to be realized. In the heterogeneous wireless multimedia sensor network, it can be logically divided into three types of nodes: converging node, source node and intermediate node. This paper presents the implementation of the WDQS algorithm on three nodes, and obtains the acquisition calculation. The scheme of various parameters of the method realizes the automatic configuration of the parameters, which provides the possibility for the algorithm from the theoretical plan to the concrete realization.
【学位授予单位】：华南理工大学
【学位级别】：博士
【学位授予年份】：2014
【分类号】：TP212.9;TN919.8

【参考文献】