# 电子工程代写|嵌入式网络系统代写Embedded Networked Systems代考|SE5201

## 电子工程代写|嵌入式网络系统代写Embedded Networked Systems代考|Genetic Algorithms

Palesi and Givargis [1] propose a method for exploring design space for a systemon-a-chip $(\mathrm{SoC})$ architecture using a Genetic Algorithm (GA). They assume that the architecture is dependent on a set of parameters which have a large effect on the performance of the application run on the chip, but is difficult to tune manually. In order to run a GA it is necessary to define three things: A representation of the configuration (gene), an objective function or goodness measure and thirdly a convergence criterion which defines when a solution is found. In this example, and for any reconfiguration task, the represented gene is a vector of the parameters of the system. The function is in this case the power consumption and the time required to perform the task to be implemented. Finally the convergence criterion is a maximum number of iterations or a Pareto-optimal solution.

What a genetic algorithm then will do is iteratively try multiple variations of the configuration, and determine what the utilities (or fitness) of the proposed configurations are [2]. A selection is made from the top scoring configurations, and alterations are made upon these configurations for the next iteration. This method has some very clear links with the Darwinist theory of evolution. The exact method of altering existing configurations may vary from mutations where any parameter changes by a random offset, or crossover where complete sections are copied from other viable solutions. Some variants even allow for invaders which are completely random new competitors added to the pool at each iteration to guarantee a larger coverage in the search space area.

Using genetic algorithms has some advantages, namely that they are generic; little to no knowledge about the problem is required except for the three abovementioned things. Another advantage is that GA algorithms satisfy the anytime algorithm criterion: it will always come up with a solution, but the refinement improves as the algorithm runs for a longer period of time.

## 电子工程代写|嵌入式网络系统代写Embedded Networked Systems代考|Graph Based Methods

Georgas [4] already proposed a method for describing reconfiguration as graph based models. This method called ARCM (Architecture Runtime Configuration Management) is specifically aimed to improve the visibility and understandability of runtime adaptive processes while allowing for human input in the adaptation-control loop.
At the core of their method (or linchpin as the authors put it) is the architectural configuration graph. Each node in this graph is a configuration, and the edges are transitions with a specific set of conditions. Whenever something is changed in the configuration of the system, this is stored in a graphical representation, and the amount of times the configuration is chosen, as well as the amount of time spent in that configuration is stored. Afterwards this graph can be inspected by the user in order to see how the system (re)acted.

Finite state machines have been used to describe reconfiguration by Teich and Köster [5]. They introduce a concept of self-reconfigurable finite state machines where in iterative steps a finite state machine can reconfigure itself by changing at most one transitions/output pair. In order to change a complete chain, some intermediate temporary configurations are required. If each intermediate configuration is considered a city, then the problem of reconfiguration in this case becomes comparable to the traveling salesman problem, and no algorithm can find an optimal way to reconfigure in polynomial time. Therefore the authors use a Genetic Algorithm approach to find a way to solve the reconfiguration problem. It should be noted that in this case the actual application that is reconfigured is a finite state machine, whereas the reconfiguration problem is not a finite state machine.

## 电子工程代写|嵌入式网络系统代写Embedded Networked Systems代考|Genetic Algorithms

Palesi 和 Givargis [1] 提出了一种探索片上系统设计空间的方法(小号○C)使用遗传算法 (GA) 的架构。他们假设架构依赖于一组参数，这些参数对芯片上运行的应用程序的性能有很大影响，但很难手动调整。为了运行遗传算法，有必要定义三件事：配置（基因）的表示，目标函数或优度度量，第三是定义何时找到解决方案的收敛标准。在此示例中，对于任何重新配置​​任务，表示的基因是系统参数的向量。在这种情况下，函数是执行要执行的任务所需的功耗和时间。最后，收敛标准是最大迭代次数或帕累托最优解。

## 电子工程代写|嵌入式网络系统代写Embedded Networked Systems代考|Graph Based Methods

Georgas [4] 已经提出了一种将重构描述为基于图的模型的方法。这种称为 ARCM（架构运行时配置管理）的方法专门用于提高运行时自适应过程的可见性和可理解性，同时允许在自适应控制循环中进行人工输入。

Teich 和 Köster [5] 使用有限状态机来描述重新配置。他们引入了自重构有限状态机的概念，其中在迭代步骤中，有限状态机可以通过最多更改一个转换/输出对来重新配置自己。为了改变一条完整的链，需要一些中间临时配置。如果将每个中间配置视为一个城市，那么这种情况下的重构问题就可以与旅行商问题相媲美，并且没有算法可以在多项式时间内找到最优的重构方式。因此，作者使用遗传算法方法来寻找解决重新配置问题的方法。应该注意的是，在这种情况下，重新配置的实际应用程序是有限状态机，

myassignments-help数学代考价格说明

1、客户需提供物理代考的网址，相关账户，以及课程名称，Textbook等相关资料~客服会根据作业数量和持续时间给您定价~使收费透明，让您清楚的知道您的钱花在什么地方。

2、数学代写一般每篇报价约为600—1000rmb，费用根据持续时间、周作业量、成绩要求有所浮动(持续时间越长约便宜、周作业量越多约贵、成绩要求越高越贵)，报价后价格觉得合适，可以先付一周的款，我们帮你试做，满意后再继续，遇到Fail全额退款。

3、myassignments-help公司所有MATH作业代写服务支持付半款，全款，周付款，周付款一方面方便大家查阅自己的分数，一方面也方便大家资金周转，注意:每周固定周一时先预付下周的定金，不付定金不予继续做。物理代写一次性付清打9.5折。

Math作业代写、数学代写常见问题

myassignments-help擅长领域包含但不是全部: