NS2模拟Ping

http://www.isi.edu/nsnam/ns/tutorial/跟着这个添加自己的ping协议，环境是Ubuntu10.04, NS2.34,其实这个版本的NS已经实现的ping协议，并且与前面网站版本不同，在改代码的时候会有所不同。首先添加自己的ping.h头文件：

#ifndef ns_ping_h
#define ns_ping_h

#include "agent.h"
#include "tclcl.h"
#include "packet.h"
#include "address.h"
#include "ip.h"

struct hdr_ping {
    //set '0' if the packet is on its way from the sender to the node which is being pinged,while it is going to
//be set to '1'on its way back.
    char ret;
    //a time stamp that is set on the packet when it is sent.
    double send_time;
}

class PingAgent : public Agent {
    public :
        PingAgent();
        int command(int argc, const char * const * argv);
        void recv(Packet *, Handler *);
    protected :
        int off_ping_;
}
#endif

注意为了不影响已经存在的分组头，最好是自己添加命名规则一般是hdr_new。

static class PingHeaderClass : public PacketHeaderClass {
    public :
        PingHeaderClass() : PacketHeaderClass("PacketHeader/Ping",
                    sizeof(hdr_ping)) {}
}class_pinghdr;

static class PingClass : public TclClass {
    public :
        PingClass() : TclClass("Agent/Ping") {}
        TclObject *create(int, const char * const *) {
            return (new PingAgent());
        }
}class_ping;

PingAgent :: PingAgent() : Agent(PT_PING)
{
    bind("packetSize_", &size);
    bind("off_ping_", &off_ping_);
}

int PingAgent :: command(int argc, const *char *const argv)
{
    if(argc == 2) {
        if(strcmp(argv[1], "send") == 0) {
            //create a new packet
            Packet* pkt = allocpkt();
            //access the ping header for tne new packet
            hdr_ping* hdr = (hdr_ping*)pkt->access(off_ping_);
            //set the 'ret'    field to 0, so the receiving node knows
//that it has to generate an echo packet
            hdr->ret = 0;
            //store the current time in the 'send_time' field
            hdr->send_time = Schedulet::instance.clock();
            //send the packet
            send(pkt, 0);
            //return TCL_OK, so the calling function knows that 
//the command has been processed
            return TCL_OK;
        }
    }

    //if the command hasn't been processed by PingAgent()::command,
//call the command() function for the base class
    return (Agent::command(argc, argv));
}


void PingAgent :: recv(Pakcet *pkt, Handler *)
{
    //aceess the IP header for the recived packet:
    hdr_ip * hdrip = (hdr_ip*)pkt->access(off_ip_);
    //access the ping header for the recived packet:
    hdr_ping * hdr = (hdr_ping*)pkt->access(off_ping_);

    if(hdr->ret == 0) {
        //send an 'echo'.First save the old packet's send_time/
        double stime = hdr->send_time;
        //discard the packet
        Packet::free(pkt);
        //create a new packet
        Packet * pktret = allocpkt();
        //access the ping header for the new packet
        hdr_ping* hdrret = (hdr_ping*)pktret->access(off_ping_);
        //set the 'ret' field to 1, so the receiver won't send another echo 
        hdrret->ret = 1;
        //set the send time field to the correct value
        hdrret->send_time = stime;
        //send the packet
        send(pktret, 0);
    } else {
        //a packet was received. Use tcl.eval to call the Tcl
//interreter with the ping results.
//note : In the Tcl code, a procedure 'Agent/Ping recv {form rtt}'
//has to be defined which allows the user to react to the ping
//result.
        char out[100];
        //prepare the output to the Tcl interpreter. Calculate the round
//trip time.
        sprintf(out, "%s recv %d %3.1f", name(), 
                    hdrip->src_.addr_ >> Address::instance().NodeShitf_[1],
                    (Scheduler::instance().clock()-hdr->send_time) * 1000);
        Tcl& tcl = Tcl::instance();
        tcl.eval(out);
        //discard the packet
        Packet::free(pkt);
    }
}

有4个C++类与分组头和分组有关，它们分别是Packet, p_info, PacketHeaderClass和PacketHeaderManager.其中 PacketHeaderClass是各种分组头的基类，而它本身是继承于TclClass的，现在对PingHeaderClass的作用理解不是很清楚，而PingClass用于创建一个PingAgent对象，注意关键字static，这样保证了当NS2启动时能调用create返回一个TclObject对象（NS2的分裂对象模型），PingAgent的构造函数绑定了两个变量。command方法牵涉到cmd机制，cmd命令是meet the Tcl Unknown mechanism——Tcl的unknown机制，简单来说，当向ns输入一个未知的命令时，Tcl的unknow机制就会调用cmd函数，cmd函数激活shadow对象的command方法，进而执行这个command方法，在这里我们在command中添加了send命令。在这里我们并没有重写父类Agent的send方法，可以在~ns/common/agent.{cc, h}中看到send的实现：

void send(Packet* p, Handler* h) { target_->recv(p, h); }

所以，send方法调用了接受者的recv方法，而我们实现了recv方法，并调用tcl对象的recv方法，这就是为什么在下面的tcl文件中recv方法被调用的理由。

当然，还需要一些必要的改变：

首先要增加分组头类型，在NS2.34中，修改了packet_t的类型，在ns2.33之前packet_t是enum，而之后定义为unsigned int.所以添加如下：

static const packet_t PT_PING = 44;

需要注意的是，ns2.34已经实现了ping协议，所以我并没有添加任何内容，如果是自己另写协议，在添加packet_t时，必须注释掉不用的分组头，并且要按所有常量值的顺序存储，不然就会有如下错误：

diffserv -I./satellite -I./wpan -I./mit/rca -I./mit/uAMPS -o common/ptypes2tcl.o common/ptypes2tcl.cc
g++ -Wl,-export-dynamic  -o common/ptypes2tcl common/ptypes2tcl.o
./common/ptypes2tcl > gen/ptypes.cc
Segmentation fault
make: *** [gen/ptypes.cc] Error 139

PS：还需要该/bin目录的权限：

chmod -R 755 /bin

不然会有权限问题，还得用sudo  make。

在p_info类中添加内容时也可网站不同：

p_info()
     {
         initName();
     }

所以需要在initName()方法中添加名字。另外，需要在~ns/tcl/lib/ns-default.tcl中添加包的大小：

Agent/Ping set packetSize_ 64

然后在~ns/tcl/lib/ns-packet.tcl的foreach prot模块中添加：

Ping    # Ping

最后在makefile文件中添加，直接复制源码吧（自己以后注意添加就是了）：

tcp/tfrc.o tcp/tfrc-sink.o mobile/energy-model.o apps/ping.o tcp/tcp-rfc793edu.o

可以看到，ns自带的ping源码在apps/目录下。

再加上tcl文件：

#create a simulator object
set ns [new Simulator]

#open a trace file
set tracefd [open out.tr w]
$ns trace-all $tracefd
set nf [open out.nam w]
$ns namtrace-all $nf

#define a finish procedure
proc finish {} {
    global ns nf
    $ns flush-trace
    close $nf
    exec nam out.nam &
    exit 0
}

#create three nodes
set n0 [$ns node]
set n1 [$ns node]
set n2 [$ns node]

#connect the nodes with two links
$ns duplex-link $n0 $n1 1Mb 10ms DropTail
$ns duplex-link $n1 $n2 1Mb 10ms DropTail

#define a recv funciton for the class Agent/Ping
Agent/Ping instproc recv {from rtt} {
    $self instvar node_
    puts "node [$node_ id] received ping answer from/
        $from with round-trip-time $rtt ms."
    }

#create two ping agents and attach them to the nodes n0 and n2
set p0 [new Agent/Ping]
$ns attach-agent $n0 $p0

set p1 [new Agent/Ping]
$ns attach-agent $n2 $p1

#connect
$ns connect $p0 $p1

$ns at 0.2 "$p0 send"
$ns at 0.4 "$p1 send"
$ns at 0.6 "$p0 send"
$ns at 0.6 "$p1 send"
$ns at 1.0 "finish"

$ns run

可以看到在tcl文件中，调用了send命令，并且添加了tcl对象的recv方法，方便c++shadow对象调用。需要注意的是，connect方法并没有建立连接，只是告知目的地址，并且在ns中，应用层之间并没有发送数据，都是由代理模拟的。

疑问：对NS2的内部运行机制还是半懂不懂，并且不理解单线程是怎么实现多节点同时发送数据的，另外，对PacketHeaderManager类的作用也还没有体会到。