✅作者简介：热爱科研的Matlab仿真开发者，修心和技术同步精进，matlab项目合作可私信。
🍎个人主页：Matlab科研工作室
🍊个人信条：格物致知。
更多Matlab仿真内容点击👇
⛄ 内容介绍
Distributed systems such as Grid- and Cloud Computing provision web services to their users in all of the world. One of the most important concerns which service providers encounter is to handle total cost of ownership (TCO). The large part of TCO is related to power consumption due to inefficient resource management. Task scheduling module as a key component can has drastic impact on both user response time and underlying resource utilization. Such heterogeneous distributed systems have interconnected different processors with different speed and architecture. Also, the user application which is typically presented in the form of directed acyclic graph (DAG) must be executed on this type of parallel processing systems. Since task scheduling in such complicated systems belongs to NP-hard problems, existing heuristic approaches are no longer efficient. Therefore, the trend is to apply hybrid metaheuristic approaches. In this paper, we extend a meta-heuristic shuffled genetic-based task scheduling algorithm to minimize total execution time, makespan, of user application. In this regard, we take benefit of other heuristics such as Heterogeneous Earliest Finish Time (HEFT) approach to generate smart initial population by applying a new shuffle operator which makes a fortune to explore feasible and promising individuals in the search space. We also conduct other genetic operators in right way to produce final near to optimal solution. To reach concrete results we have conducted several scenarios. Our proposed algorithm outperforms in term of average makespan compared with other existing approaches such as HEFT versions and QGARAR.
⛄ 部分代码
%% Shuffled Genetic Algorithm for Task Scheduling based on:
% Hosseini, M. (2018). A new Shuffled Genetic-based Task
%Scheduling Algorithm in Heterogeneous Distributed Systems.
% Journal of Advances in Computer Research, 9(4), 19-36.
clc
clear
close all
%%
global nVM nTask DAG extP0 extP1 extP2 c
nVM=3; % Number of Hetergenous Virtual Machines
c=ones(nVM)-eye(nVM); % communication time between servers
nTask=11; % Number of Tasks
DAG=[0 12 14 0 0 0 0 0 0 0 0 % Directed Acyclic Graph
0 0 0 8 15 11 0 0 0 0 0
0 0 0 0 0 0 13 0 0 0 0
0 0 0 0 0 0 0 11 0 0 0
0 0 0 0 0 0 0 8 0 0 0
0 0 0 0 0 0 0 0 7 12 0
0 0 0 0 0 0 0 0 0 14 0
0 0 0 0 0 0 0 0 0 0 15
0 0 0 0 0 0 0 0 0 0 7
0 0 0 0 0 0 0 0 0 0 10
0 0 0 0 0 0 0 0 0 0 0];
extP0=[7 10 5 6 10 11 12 10 8 15 8]; %Execution Time on Processor1
extP1=[9 9 7 8 8 13 15 13 9 11 9];
extP2=[8 14 6 7 6 15 18 7 10 13 10];
Wbar=[8 11 6 7 8 13 15 10 9 13 9]; % Average Computation Cost
npop=20; % population size
maxIter= 100; % maximum number of generation
%% The first generation:
population=INITp(npop);
%% Genetic optimization:
for iter=1:maxIter
for i=1:npop
cost(i)=MAPPER(population{i});
fitness(i)=1/cost(i);
end
for i=1:npop
probs(i)=fitness(i)/sum(fitness);
end
[val,idx]=sort(cost);
best=population{idx(1)};
min_cost=val(1);
disp(['Generation ',num2str(iter),' ... min cost= ',num2str(min_cost)]);
plot(iter,min_cost,'ko');
hold on
pause(0.000001)
for i=1:npop
if i<= 0.2 * npop
newPopulation{i}=population{idx(i)};
else
id1=randsrc(1,1,[1:npop;probs]); % Roullete wheel
id2=randsrc(1,1,[1:npop;probs]);
parent1=population{id1};
parent2=population{id2};
newPopulation{i}=crossover_mutation(parent1,parent2);
end
end
population=newPopulation;
end
final_solution=best
makespan_genetic=val(1)
makespan_Level=MAPPER(1:11)
makespan_rankd=MAPPER([1 2 3 4 7 6 5 8 9 10 11])
makespan_ranku=MAPPER([1 3 2 7 6 4 5 8 10 9 11])
⛄ 运行结果
