收集的一些算法代码，包括遗传算法、粒子群算法、蚁群算法matlab程序
选播是Internet中一种新型的网络服务,是IPv6的新特性;它使用户通过一个选播地址就能访问到该地址所表示的一组服务器主机中对用户来说“最近”的一个。随着计算机网络技术的快速发展,音频、视频等流媒体业务对网络服务质量(QoS)提出了更高的要求。如何在选播网络中实现多种服务质量已经成为了人们的研究热点。目前涌现的很多选播路由算法因为网络具有时延、时延抖动和丢包率等特点,无法满足实时业务需求。针对这个问题,本文对QoS选播路由进行了研究。提出了一种基于禁忌搜索负载均衡QoS选播路由算法,该算法针对网络节点数越大传输服务质量越差的缺点,引入权值变换函数,将有带宽、时延约束的二维权值转化为一维权值,简化了目标函数求解的复杂性,保证了搜索的准确性。并且在搜索过程中,将网络的代价和负载均衡度作为目标函数,使路由所需费用较小而负载均衡分布较好,从而提高了网络资源的利用效率,降低了网络成本,改善网络整体性能。提出了一种基于混合遗传多约束QoS选播路由算法,该算法针对遗传算法在生物进化后期适应度值相差不明显的情况,采用模拟退火算法局部收敛高的优势对其适应度进行拉升,使生物个体后期的适应度值又有较大的差距,从而使优良个体突出得以选出和遗传。该算法加快了遗传算法收敛速度和提高了遗传算法局部收敛的性能,具有良好的求解效果,可以找到满足时延、时延抖动要求的最小费用的路由选播树。通过实验仿真结果表明,第一种算法求得路径的时延与Anycast-Routing算法差不多,但它能够较好地平衡网络资源的利用效率,改善网络性能,降低网络成本。第二种算法求得的进化代数与MDA算法的进化代数也相差不大,可以满足多约束QoS选播的请求,达到全局收敛。因此,两种算法都是有效的、可行的。
本文的目的就是针对BP算法的这些局限性,如BP算法的学习速率较慢,网络会产生部分平坦区,网络权重的误差变化量在平坦区的变化几乎为零,致使整个网络的训练过程停止等,通过对算法结构及原理的分析,提出新的改进算法,使新算法较原算法在收敛速度和训练次数等方面有显著的提高。文中分别从标准权重更新公式中加入新的比例因子,并将训练中为常数的学习速率调整为随着训练逐渐改变的量,利用增益变化项可以修改本地梯度的同时给每次的迭代训练提供一个改进的梯度搜索方向,及基于拟牛顿法的Levenberg&Marquardt算法三个方向分别改进。在反向传播算法中,学习速率的大小直接影响着整个网络的收敛速度。因为学习速率和梯度是权重连接数变化量的直接影响因素。但是,在标准BP算法中,学习速率在网络训练开始前已经初始化为一个常数。固定学习率的梯度下降法是低效的。因此提出了一种新的高效的方法,让学习速率适时调整,即第K次网络训练的学习速率与第K-1次的学习速率有关,并在权值变化公式中在原来两个影响因子(动量因子和学习速率因子)的基础上加入一个新的比例因子来调节每一次迭代所得到的输出和目标的差异,使网络的训练更准确,并且改善
Anycast is a new
kind of network services in the Internet, and it is a new feature
of IPv6. Through anycast address, users access to the closest
server host of sets, which are denoted by this anycast address.
With the rapid development of network technology, some businesses
about audio, video and other streaming media have higher
requirement for quality of service (QoS). How
Anycast is a new
kind of network services in the Internet, and it is a new feature
of IPv6. Through anycast address, users access to the closest
server host of sets, which are denoted by this anycast address.
With the rapid development of network technology, some businesses
about audio, video and other streaming media have higher
requirement for quality of service (QoS). How to implement many QoS
in anycast networks has becomes a research hotspot.The current
anycast routing algorithms cann’t meet the demand for real-time
business for characteristic of network, such as delay, delay jitter
and packet loss rate, etc. Therefore, the researches in
Anycast is a new kind
of network services in the Internet, and it is a new feature of
IPv6. Through anycast address, users access to the closest server
host of sets, which are denoted by this anycast address. With the
rapid development of network technology, some businesses about
audio, video and other streaming media have higher requirement for
quality of service (QoS). How to implement many QoS in anycast
networks has becomes a research hotspot.The current anycast routing
algorithms cann’t meet the demand for real-time business for
characteristic of network, such as delay, delay jitter and packet
loss rate, etc. Therefore, the researches in this thesis focus on
the anycast route.Anycast QoS routing algorithm balancing Network
load based tabu search is proposed. In view of the more network
nodes, the worse QoS will be, the algorithm import value transform
function. Which transform two dimensions values with bandwidth and
delay-constrained into one dimension value. That simplifies the
complexity of solving object function and improves searching
precision. During searching, the algorithm use value of network and
load balancing as object function, so that routing cost is less and
the distribution of load balancing is better, thus enhancing the
efficient of network resources, reducing network costs, and improve
network performance.Anycast routing algorithm with solution of
multiple QoS constraints bases on hybrid genetic algorithm(HGA) is
proposed. According to the difference of adaptive degree is not
obvious in HGA’s latter part of biological evolution, the algorithm
adopt simulated annealing algorithem to converge high advantage,
and improve adaptive degree, in order to increase the difference of
adaptive degree in biological individual latter. Therefore the
algorithm can select excellent individual to inherit. This
algorithm speed up the convergence rate and improve local
convergence performance of genetic algorithm, and has a good
solution effects. It can find minimum cost anycast routing tree
which meets the delay, and delay jitter requirements.Simulation
results prove that the delay which the first algorithm calculated
is almost same as Anycast-Routing, but the first algorithm can
balance the efficient of network resources, improve network
performance and reduce network cost. Evolutional algebra which the
Second algorithm calculated is also same as MDA the
second algorithm can meet the request of multi-constrained Qos
anycast, in order to entirely convergence. Therefore, the two
algorithms are effective and feasible.
已投稿到：
以上网友发言只代表其个人观点，不代表新浪网的观点或立场。(window.slotbydup=window.slotbydup || []).push({
id: '2014386',
container: s,
size: '234,60',
display: 'inlay-fix'
&&|&&0次下载&&|&&总48页&&|
您的计算机尚未安装Flash，点击安装&
阅读已结束，如需下载到电脑，请使用积分（）
下载：20积分
2人评价38页
0人评价19页
3人评价11页
1人评价66页
7人评价149页
所需积分：（友情提示：大部分文档均可免费预览！下载之前请务必先预览阅读，以免误下载造成积分浪费！）
（多个标签用逗号分隔）
文不对题，内容与标题介绍不符
广告内容或内容过于简单
文档乱码或无法正常显示
若此文档涉嫌侵害了您的权利，请参照说明。
评价文档：
下载：20积分 上传我的文档
 下载
 收藏
该文档贡献者很忙，什么也没留下。
 下载此文档
正在努力加载中...
模拟退火法与禁忌搜索法求解vrp的对比分析
下载积分：2000
内容提示：模拟退火法与禁忌搜索法求解vrp的对比分析
文档格式：PDF|
浏览次数：40|
上传日期： 08:58:46|
文档星级：
全文阅读已结束，如果下载本文需要使用
 2000 积分
下载此文档
该用户还上传了这些文档
模拟退火法与禁忌搜索法求解vrp的对比分析
官方公共微信