void network_status(){
dhcp_pair *pair = NULL;
pair = (dhcp_pair *) malloc(sizeof(dhcp_pair));
memset(pair, 0, sizeof(dhcp_pair));
// Read from /tmp/wista.txt or from command to get mac link to server
read_wlan_client(pair);
// Form json to get ip
char *json = form_request_json(pair);
// Send Request
char response[4096];
int resp_len;
memset(response, 0, sizeof(response));
//send_request("192.168.48.1", PORT, "/dhcplease", json， strlen(json), response, resp_len);
send_request("127.0.0.1", PORT, "/dhcplease", json, strlen(json), response, resp_len);
//printf("response[%d]=[%s]\n", resp_len, response);
char *p = strstr(response, "\r\n\r\n");
if (p) {
parse_response_json(p+strlen("\r\n\r\n"), pair);
}
result_array ret[64];
memset(&ret, 0, sizeof(ret));
int i=0;
for(i=0; i<pair->length; i++) {
//printf("mac[%d]=%s\n", i, pair->value[i].mac);
//printf(" ip[%d]=%s\n", i, pair->value[i].ip);
ret[i] = network_delay(pair->value[i].ip);
int j=0;
for (j=0; j< PKT_LEN; j++)
printf("ip[%s]-max[%f]-min[%f]-avg[%f]\n", pair->value[i].ip,
ret[i].icmp[j].max, ret[i].icmp[j].min, ret[i].icmp[j].avg);
}
//TODO: No json defined
//Form json post to magicwifi
//Send request to magicwifi
}
int main()
{
srand((unsigned)time(NULL));
string address="graph_all.txt";
ofstream outfile("result.txt");//实验结果
ofstream req_outfile("reqs.txt");//实验用的流需求
outfile<<address<<"网络拓扑"<<" TESTNUM="<<TESTNUM<<endl;
outfile<<"APPNUM="<<APPNUM<<endl;
outfile<<"Maxreq="<<Maxreq<<endl;
outfile<<"caseN: "<<caseN<<endl;
outfile<<"flow Range: "<<Begin_num<<"--"<<Maxflow+Begin_num-1<<endl<<endl;
//统计所有case结果
vector<double> oneCaseHappiness_Voting;//Voting结果的单轮满意度
vector<double> multiCaseHappiness_Voting;//Voting结果的多轮满意度
vector<double> oneCaseS2_Voting;//Voting结果的单轮满意度方差
vector<double> delay_Voting;//Voting结果的延时
vector<double> TE_Voting;////Voting结果的最大链路利用率
vector<double> oneCaseHappiness_TENetwork;//以TE为目标集中优化结果的单轮满意度
vector<double> multiCaseHappiness_TENetwork;
vector<double> delay_TENetwork;
vector<double> TE_TENetwork;
vector<double> oneCaseHappiness_DelayNetwork;//以Delay为目标集中优化结果的单轮满意度
vector<double> multiCaseHappines_DelayNetwork;
vector<double> oneCaseS2_DelayNetwork;
vector<double> delay_DelayNetwork;
vector<double> TE_DelayNetwork;
vector<double> s2_DelayNetwork_Voting;//DelayNetwork的方差比上投票的结果
for(int test=0;test<TESTNUM;test++)
{
//outfile<<"********************* test "<<test<<"*********************"<<endl;
cout <<"********************* test "<<test<<"*********************"<<endl;
vector<APP*> appL;//记录所有的APP
vector<Req*> reqL;//流需求
VGraph gv(address);//Voting用的图
TENetworkGraph gn_te(address);//TE全局优化的图
DelayNetworkGraph gn_delay(address);//延时全局优化的图
double happiness_sum=0;//voting的多轮满意度
double happiness_sum_teLP=0;//te网络的多轮满意度
double happiness_sum_delayLP=0;//delay网络的多轮满意度
double table[M2C+1][N2C+1] = {0};//投票用的输入
int ranking[N2C+1]={0};//记录一种排序的投票人数
for(int k=0;k<APPNUM;k++)
appL.push_back(new APP(k));
for(int j=0;j<Maxreq;j++)
ranking[j]=1;//每种投票结果有1个voter,如果为2就说明该方案有得到两个voter的票
//每个case是一波流(同时到达的Maxreq条流)
for(int i=0;i<caseN;i++)
{
cout<<endl<<"*************************"<<" case "<<i<<"*************************"<<endl;
//初始化
for(int j=0;j<Maxreq;j++)
for(int k=0;k<Maxreq;k++)
table[j][k]=0;
reqL.clear();
gv.reqL.clear();
for(int j=0;j<APPNUM;j++)
appL[j]->init();
//以下程序时防止每个case都会创建的Req导致的内存泄露
vector<Req*>::iterator it,iend;
iend=reqL.end();
for(it=reqL.begin();it!=iend;it++)
delete(*it);
//随机产生流量: (id app_id src dst flow)
req_outfile<<"case "<<i<<endl;
for(int j=0;j<Maxreq;j++)
{
int app_id=0,a=0,b=0,c=0;
while(1){
c = Begin_num + rand()%Maxflow;
a = rand()%N;
b = rand()%N;
app_id=rand()%APPNUM;
if(a!=b && c!=0) break;
}
//if(j==0){app_id=j;a=0;b=3;c=5;}
//else if(j==1){app_id=j;a=1;b=3;c=5;}
//else if(j==2){app_id=j;a=0;b=1;c=5;}
//app_id=j;//一个app一条流
Req* r = new Req(j,app_id,a,b,c);
reqL.push_back(r);
gv.reqL.push_back(r);
req_outfile<<j<<" "<<app_id<<" "<<a<<" "<<b<<" "<<c<<endl;
}
//@@@@@@@@@@@@@@@@ Voting @@@@@@@@@@@@@@@@@@@@@@
//@@@@@@@@@@@@@@@@ Voting @@@@@@@@@@@@@@@@@@@@@@
//算最完美方案，所有的路都第一个走得到的方案
for(int j=0;j<APPNUM;j++)
{
vector<Req*> reqPL_app;
for(int i=0;i<reqL.size();i++)
{
if(j==reqL[i]->app_id) reqPL_app.push_back(reqL[i]);
}
gv.network_delay(reqPL_app,j);
}
//for(int j=0;j<Maxreq;j++)
//cout<<"gv.cost_best "<<j<<" : "<<gv.cost_best[j]<<endl;
//提方案
for(int j=0;j<APPNUM;j++){
cout<<"app "<<j<<" proposal"<<endl;
appL[j]->propose(gv,appL);
}
//评价方案
for(int j=0;j<APPNUM;j++)
appL[j]->evaluate(gv,appL);
//投票算法
for(int j=0;j<APPNUM;j++)
for(int k=0;k<APPNUM;k++)
{
if(appL[j]->judge[k]==0) table[j][k]=10000;//如果是0，说明流没有摆在网络中
else table[j][k]=appL[j]->judge[k];
}
cout<<endl<<"voting uses ";
int choice=1;//选择一种投票算法
int winner=0;
Voting vv(table,ranking,APPNUM,APPNUM);
winner=vv.voting(choice);
//**************************** 最高满意度方案和最小满意度方差方案 评价
//计算 最高满意度方案 和 最小满意度方差方案
double happinessVotingMax=0;//最高满意度
double happinessVotingTemp=0;
int happinessVotingLoc=0;
double s2Min=Inf;//最小满意度方差
double s2Temp=0;
int s2Loc=0;
double s2Min_Happiness=0;//最小满意度方差方案的满意度
for(int k=0;k<APPNUM;k++)
{
happinessVotingTemp=0;
for(int j=0;j<APPNUM;j++)
happinessVotingTemp+= gv.cost_best[j]/table[j][k];//最好抉择评分/当前抉择评分
if(happinessVotingMax<happinessVotingTemp/APPNUM){
happinessVotingMax=happinessVotingTemp/APPNUM;
happinessVotingLoc=k;
}
s2Temp=0;
for(int j=0;j<APPNUM;j++)
s2Temp+= (gv.cost_best[j]/table[j][k]-happinessVotingTemp/APPNUM) * (gv.cost_best[j]/table[j][k]-happinessVotingTemp/APPNUM);
if(s2Min>s2Temp/APPNUM){
s2Min=s2Temp/APPNUM;
s2Loc=k;
s2Min_Happiness=happinessVotingTemp/APPNUM;
}
}
//计算 最高满意度方案 的 应用满意度方差
double s2VotingMax=0;
//最高满意度方案的应用满意度的平均值
double happinessVotingMaxAvg=happinessVotingMax/APPNUM;
for(int j=0;j<APPNUM;j++){
if(gv.cost_best[j]==0)continue;//没有流量需求的APP就没有满意度的区别
s2VotingMax+=(gv.cost_best[j]/table[j][happinessVotingLoc]-happinessVotingMaxAvg) * (gv.cost_best[j]/table[j][happinessVotingLoc]-happinessVotingMaxAvg);
}
s2VotingMax=s2VotingMax/APPNUM;
cout <<"最高满意度方案的结果"<<endl;
cout << "第" << i << "轮整体满意度： " << happinessVotingMax << endl;
cout << "第" << i << "轮满意度满意度方差： " << s2VotingMax << endl;
cout <<endl;
cout <<"最小满意度 方差 方案的结果"<<endl;
cout << "第" << i << "轮整体满意度： " << s2Min_Happiness << endl;
cout << "第" << i << "轮满意度满意度方差： " << s2Min << endl;
cout <<endl;
//**************************** END OF 最高满意度方案和最小满意度方差方案 评价
//如果选择方差最小的方案对满意度损伤超过了30%，就采用schulze method的投票结果
if(happinessVotingMax-s2Min_Happiness<=0.3) winner=s2Loc;
cout<<"crana voting method : " << winner << endl;
/*
if (choice == 1)
cout << "schulze method : " << winner << endl;
else if (choice == 2)
cout << "comulative method: " << winner << endl;
else if (choice == 3)
cout << "copeland condorcet method: " << winner << endl;
else
cout << "ranked pairs method: " << winner << endl;
*/
// table show
for(int i=0;i<APPNUM;i++)
{
cout<<"app ";
cout.setf(ios::right);
cout.width(3);
cout<<i<<" judge: ";
for(int j=0;j<APPNUM;j++){
cout.setf(ios::left);
cout.width(5);
cout<<table[i][j]<<" ";
}
cout<<endl;
}
cout<<endl;
//计算投票winner满意度
double happiness=0;//一轮所有流的满意度和，越高越好,0<=满意度<=1
for(int j=0;j<APPNUM;j++)
happiness += gv.cost_best[j]/table[j][winner];//最好抉择评分/当前抉择评分
happiness_sum += happiness;
//计算投票winner的应用满意度方差
double s2_voting=0;
double happiness_avg=happiness/APPNUM;
for(int j=0;j<APPNUM;j++)
{
if(gv.cost_best[j]==0) continue;
s2_voting+=(gv.cost_best[j]/table[j][winner]-happiness_avg) * (gv.cost_best[j]/table[j][winner]-happiness_avg);
}
s2_voting=s2_voting/APPNUM;
//统计网络latency
double latencyVoting=0;
latencyVoting=judge_sum_function(gv,appL,winner);
cout <<"Voting Winner 方案的结果"<<endl;
cout << "第" << i << "轮整体满意度： " << happiness/APPNUM << endl;
cout << "第" << i << "轮满意度满意度方差： " << s2_voting << endl;
cout << "多轮整体延时和: " << latencyVoting << endl;
double maxUtil_Voting=0;
for(int j=0;j<gv.m;j++)
{
int src=gv.incL[j]->src;
int dst=gv.incL[j]->dst;
double capacity=gv.incL[j]->capacity;
if(maxUtil_Voting<(appL[winner]->adj[src][dst]/capacity))
maxUtil_Voting=appL[winner]->adj[src][dst]/capacity;
}
cout<<"最大链路利用率: "<<maxUtil_Voting<<endl;
//记录结果，方便输出
oneCaseHappiness_Voting.push_back(happiness/APPNUM);
multiCaseHappiness_Voting.push_back(happiness_sum / ((i+1)*APPNUM));
oneCaseS2_Voting.push_back(s2_voting);
delay_Voting.push_back(latencyVoting);
TE_Voting.push_back(maxUtil_Voting);
//胜利的方案部署到所有应用的adj里
appL[winner]->end_implement(gv,appL);
cout<<endl;
//@@@@@@@@@@@@@@@@@End of Voting@@@@@@@@@@@@@@@@@@@@@@@@@
//@@@@@@@@@@@@@@@@@End of Voting@@@@@@@@@@@@@@@@@@@@@@@@@
//@@@@@@@@@@@@@@@@@Begin of network_te@@@@@@@@@@@@@@@@@@@@@@@@@
//@@@@@@@@@@@@@@@@@Begin of network_te@@@@@@@@@@@@@@@@@@@@@@@@@
//TE全局优化规划部分
cout<<endl<<" network_te result "<<endl;
//最优部署计算
for(int j=0;j<Maxreq;j++)
gn_te.cost_best[j] =gn_te.dijkstra(reqL[j]->src,reqL[j]->dst,reqL[j]->flow);
//线性规划部署
double result_network_te=0;
result_network_te=network_te(&gn_te,reqL);
//计算满意度
double happiness_teLP=0;//单轮满意度统计
if(result_network_te==Inf)happiness_teLP=0;
else
{
happiness_teLP=0;
//for(int i=0;i<reqN;i++)
// cout<<g.cost_best[i]<<" "<<g.cost_LP[i]<<endl;
for(int j=0;j<Maxreq;j++)
happiness_teLP += gn_te.cost_best[j]/gn_te.cost_LP[j];
}
happiness_sum_teLP += happiness_teLP;
cout<<"单轮满意度： "<<happiness_teLP/Maxreq<<endl;
double latency_teLP=0;
latency_teLP=delay_TENetworkGraph(gn_te);
cout << "多轮整体延时和: " << latency_teLP << endl;
cout<<"最大链路利用率: "<<result_network_te<<endl;
//记录结果，方便输出
oneCaseHappiness_TENetwork.push_back(happiness_teLP/Maxreq);
multiCaseHappiness_TENetwork.push_back(happiness_sum_teLP/(Maxreq*(i+1)));
delay_TENetwork.push_back(latency_teLP);
TE_TENetwork.push_back(result_network_te);
//@@@@@@@@@@@@@@@@@End of network_te@@@@@@@@@@@@@@@@@@@@@@@@@
//@@@@@@@@@@@@@@@@@End of network_te@@@@@@@@@@@@@@@@@@@@@@@@@
//@@@@@@@@@@@@@@@@@Begin of network_delay@@@@@@@@@@@@@@@@@@@@@@@@@
//@@@@@@@@@@@@@@@@@Begin of network_delay@@@@@@@@@@@@@@@@@@@@@@@@@
//Delay全局优化规划部分
cout<<endl<<" network_delay result "<<endl;
//最优部署计算-->直接用voting的最优结果即可
for(int j=0;j<Maxreq;j++)
gn_delay.cost_best[j] =gv.cost_best[j];
//线性规划部署
double result_network_delay=0;//近似延时
result_network_delay=network_delay(&gn_delay,reqL);
//计算满意度
double happiness_delayLP=0;//单轮满意度统计
if(result_network_delay==Inf)happiness_delayLP=0;
else
{
happiness_delayLP=0;
//for(int i=0;i<reqN;i++)
// cout<<g.cost_best[i]<<" "<<g.cost_LP[i]<<endl;
for(int j=0;j<APPNUM;j++)
{
if(gv.cost_best[j]==0)continue;
happiness_delayLP += gn_delay.cost_best[j]/gn_delay.cost_LP[j];
}
}
happiness_sum_delayLP += happiness_delayLP;
double s2_delay=0;
double happiness_avg_delay=happiness_delayLP/APPNUM;
for(int j=0;j<APPNUM;j++)
{
if(gv.cost_best[j]==0) continue;
s2_delay+=(gv.cost_best[j]/gn_delay.cost_LP[j]-happiness_avg_delay) * (gv.cost_best[j]/gn_delay.cost_LP[j]-happiness_avg_delay);
}
s2_delay=s2_delay/APPNUM;
cout<<"单轮满意度： "<<happiness_delayLP/APPNUM<<endl;
cout << "第" << i << "轮满意度满意度方差： " << s2_delay << endl;
double latency_delayLP=0;
latency_delayLP=delay_DelayNetworkGraph(gn_delay);
cout << "多轮整体延时和: " << latency_delayLP << endl;
double maxUtil_DelayNetworkGraph=0;
for(int j=0;j<gn_delay.m;j++)
{
int src=gn_delay.incL[j]->src;
int dst=gn_delay.incL[j]->dst;
double capacity=gn_delay.incL[j]->capacity;
if(maxUtil_DelayNetworkGraph<(gn_delay.adj[src][dst]/capacity))
maxUtil_DelayNetworkGraph=gn_delay.adj[src][dst]/capacity;
}
cout<<"最大链路利用率: "<<maxUtil_DelayNetworkGraph<<endl;
double s2_cmp=0;
if(s2_voting==0 && s2_delay==0) s2_cmp=1;
else if(s2_voting==0 && s2_delay!=0) s2_cmp=10;//自己凭直觉假设的
else s2_cmp=s2_delay/s2_voting;
//if(s2_cmp>10) s2_cmp=10;
cout<<endl<<"Delay联合优化方差/投票方差="<<s2_cmp<<endl;
oneCaseHappiness_DelayNetwork.push_back(happiness_delayLP/APPNUM);
multiCaseHappines_DelayNetwork.push_back(happiness_sum_delayLP/(APPNUM*(i+1)));
oneCaseS2_DelayNetwork.push_back(s2_delay);
delay_DelayNetwork.push_back(latency_delayLP);
TE_DelayNetwork.push_back(maxUtil_DelayNetworkGraph);
s2_DelayNetwork_Voting.push_back(s2_cmp);
//@@@@@@@@@@@@@@@@@End of network_delay@@@@@@@@@@@@@@@@@@@@@@@@@
//@@@@@@@@@@@@@@@@@End of network_delay@@@@@@@@@@@@@@@@@@@@@@@@@
}//一个case结束
}//一个test结束
outfile<<endl;
//以下输出顺序和CRANA_Voting表格中相同
//单轮满意度 APP-Voting TENetwork DelayNetwork
outfile<<"单轮满意度 APP-Voting TENetwork DelayNetwork"<<endl;
for(int i=0;i<TESTNUM;i++)
outfile<<oneCaseHappiness_Voting[i]<<",";
outfile<<endl;
for(int i=0;i<TESTNUM;i++)
outfile<<oneCaseHappiness_TENetwork[i]<<",";
outfile<<endl;
for(int i=0;i<TESTNUM;i++)
outfile<<oneCaseHappiness_DelayNetwork[i]<<",";
outfile<<endl;
//单轮满意度方差 APP-Voting DelayNetwork DelayNetwork/APP_Voting
outfile<<"单轮满意度方差 APP-Voting DelayNetwork DelayNetwork/APP_Voting"<<endl;
for(int i=0;i<TESTNUM;i++)
outfile<<oneCaseS2_Voting[i]<<",";
outfile<<endl;
for(int i=0;i<TESTNUM;i++)
outfile<<oneCaseS2_DelayNetwork[i]<<",";
outfile<<endl;
for(int i=0;i<TESTNUM;i++)
outfile<<oneCaseS2_DelayNetwork[i]/oneCaseS2_Voting[i]<<",";
outfile<<endl;
//网络总延时 APP-Voting TENetwork DelayNetwork
outfile<<"网络总延时 APP-Voting TENetwork DelayNetwork"<<endl;
for(int i=0;i<TESTNUM;i++)
outfile<<delay_Voting[i]<<",";
outfile<<endl;
for(int i=0;i<TESTNUM;i++)
outfile<<delay_TENetwork[i]<<",";
outfile<<endl;
for(int i=0;i<TESTNUM;i++)
outfile<<delay_DelayNetwork[i]<<",";
outfile<<endl;
//最大链路利用率 APP-Voting TENetwork DelayNetwork
outfile<<"最大链路利用率 APP-Voting TENetwork DelayNetwork"<<endl;
for(int i=0;i<TESTNUM;i++)
outfile<<TE_Voting[i]<<",";
outfile<<endl;
for(int i=0;i<TESTNUM;i++)
outfile<<TE_TENetwork[i]<<",";
outfile<<endl;
for(int i=0;i<TESTNUM;i++)
outfile<<TE_DelayNetwork[i]<<",";
outfile<<endl;
getchar();
return 0;
}
