本文旨在详细解读一个 NS-3 示例程序 tcp-bbr-example.cc
,以帮助初学者理解如何在 NS-3 中进行网络仿真。本示例代码演示了如何配置和运行一个基于 TCP-BBR 协议的网络仿真场景。
这段代码演示了一个使用 TCP BBR 协议的网络仿真例子。它主要设置了一个包含发送节点、接收节点和两个中间路由器的网络拓扑,并对网络性能进行监控。
链接示意图:
+--------+ +---------+ +---------+ +--------+
| Sender |-------| Router1 |-------| Router2 |-------|Receiver|
+--------+ ^ +---------+ ^ +---------+ ^ +--------+
| | |
| edgeLink | bottleneckLink | edgeLink
| | |
1000Mbps, 5ms 10Mbps, 10ms 1000Mbps, 5ms
1. 定义跟踪函数
这些函数用于记录吞吐量、队列大小和拥塞窗口。
static void TraceThroughput(Ptr<FlowMonitor> monitor)
{
FlowMonitor::FlowStatsContainer stats = monitor->GetFlowStats();
if (!stats.empty())
{
auto itr = stats.begin();
Time curTime = Now();
throughput << curTime << " "
<< 8 * (itr->second.txBytes - prev) / ((curTime - prevTime).ToDouble(Time::US))
<< std::endl;
prevTime = curTime;
prev = itr->second.txBytes;
}
Simulator::Schedule(Seconds(0.2), &TraceThroughput, monitor);
}
void CheckQueueSize(Ptr<QueueDisc> qd)
{
uint32_t qsize = qd->GetCurrentSize().GetValue();
Simulator::Schedule(Seconds(0.2), &CheckQueueSize, qd);
queueSize << Simulator::Now().GetSeconds() << " " << qsize << std::endl;
}
static void CwndTracer(Ptr<OutputStreamWrapper> stream, uint32_t oldval, uint32_t newval)
{
*stream->GetStream() << Simulator::Now().GetSeconds() << " " << newval / 1448.0 << std::endl;
}
void TraceCwnd(uint32_t nodeId, uint32_t socketId)
{
AsciiTraceHelper ascii;
Ptr<OutputStreamWrapper> stream = ascii.CreateFileStream(dir + "/cwnd.dat");
Config::ConnectWithoutContext("/NodeList/" + std::to_string(nodeId) +
"/$ns3::TcpL4Protocol/SocketList/" +
std::to_string(socketId) + "/CongestionWindow",
MakeBoundCallback(&CwndTracer, stream));
}
2. main
函数开始,初始化参数
在这里,通过命令行参数解析设置了一些仿真参数,例如 TCP 协议类型、延迟 ACK 计数、是否启用 PCAP 文件生成和仿真停止时间等。
int main(int argc, char* argv[])
{
// Naming the output directory using local system time
time_t rawtime;
struct tm* timeinfo;
char buffer[80];
time(&rawtime);
timeinfo = localtime(&rawtime);
strftime(buffer, sizeof(buffer), "%d-%m-%Y-%I-%M-%S", timeinfo);
std::string currentTime(buffer);
std::string tcpTypeId = "TcpBbr";
std::string queueDisc = "FifoQueueDisc";
uint32_t delAckCount = 2;
bool bql = true;
bool enablePcap = false;
Time stopTime = Seconds(100);
CommandLine cmd(__FILE__);
cmd.AddValue("tcpTypeId", "Transport protocol to use: TcpNewReno, TcpBbr", tcpTypeId);
cmd.AddValue("delAckCount", "Delayed ACK count", delAckCount);
cmd.AddValue("enablePcap", "Enable/Disable pcap file generation", enablePcap);
cmd.AddValue("stopTime",
"Stop time for applications / simulation time will be stopTime + 1",
stopTime);
cmd.Parse(argc, argv);
queueDisc = std::string("ns3::") + queueDisc;
3.设置默认配置
使用 Config::SetDefault
函数设置了一些 TCP 和队列相关的默认配置。这些配置用于设置TCP协议、发送和接收缓冲区大小、初始拥塞窗口、延迟确认计数、段大小、丢弃尾部队列的最大大小,以及qdisc的最大大小。
queueDisc = std::string("ns3::") + queueDisc;
Config::SetDefault("ns3::TcpL4Protocol::SocketType", StringValue("ns3::" + tcpTypeId));
// The maximum send buffer size is set to 4194304 bytes (4MB) and the
// maximum receive buffer size is set to 6291456 bytes (6MB) in the Linux
// kernel. The same buffer sizes are used as default in this example.
Config::SetDefault("ns3::TcpSocket::SndBufSize", UintegerValue(4194304));
Config::SetDefault("ns3::TcpSocket::RcvBufSize", UintegerValue(6291456));
Config::SetDefault("ns3::TcpSocket::InitialCwnd", UintegerValue(10));
Config::SetDefault("ns3::TcpSocket::DelAckCount", UintegerValue(delAckCount));
Config::SetDefault("ns3::TcpSocket::SegmentSize", UintegerValue(1448));
Config::SetDefault("ns3::DropTailQueue<Packet>::MaxSize", QueueSizeValue(QueueSize("1p")));
Config::SetDefault(queueDisc + "::MaxSize", QueueSizeValue(QueueSize("100p")));
4. 创建节点和链路
在这里创建了发送节点、接收节点和两个中间路由器。然后使用 PointToPointHelper
设置了发送节点到路由器1、路由器1到路由器2(瓶颈链路),以及路由器2到接收节点的点对点链路。
NodeContainer sender;
NodeContainer receiver;
NodeContainer routers;
sender.Create(1);
receiver.Create(1);
routers.Create(2);
// Create the point-to-point link helpers
PointToPointHelper bottleneckLink;
bottleneckLink.SetDeviceAttribute("DataRate", StringValue("10Mbps"));
bottleneckLink.SetChannelAttribute("Delay", StringValue("10ms"));
PointToPointHelper edgeLink;
edgeLink.SetDeviceAttribute("DataRate", StringValue("1000Mbps"));
edgeLink.SetChannelAttribute("Delay", StringValue("5ms"));
// Create NetDevice containers
NetDeviceContainer senderEdge = edgeLink.Install(sender.Get(0), routers.Get(0));
NetDeviceContainer r1r2 = bottleneckLink.Install(routers.Get(0), routers.Get(1));
NetDeviceContainer receiverEdge = edgeLink.Install(routers.Get(1), receiver.Get(0));
5. 安装网络协议栈
为每个节点安装互联网协议栈,使它们能够进行网络通信。
// Install Stack
InternetStackHelper internet;
internet.Install(sender);
internet.Install(receiver);
internet.Install(routers);
6. 配置和安装qdisc
在这里配置了qdisc(Queue Discipline),如果 bql
为真,则设置动态队列限制。然后将其安装到发送节点到路由器1和路由器2到接收节点的链路上。
// Configure the root queue discipline
TrafficControlHelper tch;
tch.SetRootQueueDisc(queueDisc);
if (bql)
{
tch.SetQueueLimits("ns3::DynamicQueueLimits", "HoldTime", StringValue("1000ms"));
}
tch.Install(senderEdge);
tch.Install(receiverEdge);
7. 分配 IP 地址
分配 IP 地址给每个链路,并填充全局路由表。
// Assign IP addresses
Ipv4AddressHelper ipv4;
ipv4.SetBase("10.0.0.0", "255.255.255.0");
Ipv4InterfaceContainer i1i2 = ipv4.Assign(r1r2);
ipv4.NewNetwork();
Ipv4InterfaceContainer is1 = ipv4.Assign(senderEdge);
ipv4.NewNetwork();
Ipv4InterfaceContainer ir1 = ipv4.Assign(receiverEdge);
// Populate routing tables
Ipv4GlobalRoutingHelper::PopulateRoutingTables();
8. 设置应用程序
首先在发送端安装了一个 BulkSendHelper 应用程序,设定发送端的最大字节数为0,表示发送无限数据。然后在接收端安装一个 PacketSinkHelper 应用程序,以接收数据包。
// Select sender side port
uint16_t port = 50001;
// Install application on the sender
BulkSendHelper source("ns3::TcpSocketFactory", InetSocketAddress(ir1.GetAddress(1), port));
source.SetAttribute("MaxBytes", UintegerValue(0));
ApplicationContainer sourceApps = source.Install(sender.Get(0));
sourceApps.Start(Seconds(0.1));
// Hook trace source after application starts
Simulator::Schedule(Seconds(0.1) + MilliSeconds(1), &TraceCwnd, 0, 0);
sourceApps.Stop(stopTime);
// Install application on receiver
PacketSinkHelper sink("ns3::TcpSocketFactory", InetSocketAddress(Ipv4Address::GetAny(), port));
ApplicationContainer sinkApps = sink.Install(receiver.Get(0));
sinkApps.Start(Seconds(0.0));
sinkApps.Stop(stopTime);
9. 安装qdisc和启用跟踪
这里卸载并重新安装了qdisc,以便可以跟踪队列大小。如果启用了 PCAP 生成选项,还会生成 PCAP 文件用于后期分析。同时打开文件以写入吞吐量和队列大小的跟踪数据。
// Trace the queue occupancy on the second interface of R1
tch.Uninstall(routers.Get(0)->GetDevice(1));
QueueDiscContainer qd;
qd = tch.Install(routers.Get(0)->GetDevice(1));
Simulator::ScheduleNow(&CheckQueueSize, qd.Get(0));
// Generate PCAP traces if it is enabled
if (enablePcap)
{
MakeDirectories(dir + "pcap/");
bottleneckLink.EnablePcapAll(dir + "/pcap/bbr", true);
}
// Open files for writing throughput traces and queue size
throughput.open(dir + "/throughput.dat", std::ios::out);
queueSize.open(dir + "/queueSize.dat", std::ios::out);
NS_ASSERT_MSG(throughput.is_open(), "Throughput file was not opened correctly");
NS_ASSERT_MSG(queueSize.is_open(), "Queue size file was not opened correctly");
10. 使用 Flow Monitor 监控流量
使用 Flow Monitor 监控所有节点的流量,并调度 TraceThroughput
函数以定期记录吞吐量。
// Check for dropped packets using Flow Monitor
FlowMonitorHelper flowmon;
Ptr<FlowMonitor> monitor = flowmon.InstallAll();
Simulator::Schedule(Seconds(0 + 0.000001), &TraceThroughput, monitor);
总结
这段代码展示了如何使用 ns-3 仿真 TCP BBR 协议的性能。通过配置网络拓扑、安装应用程序和启用各种跟踪功能,可以详细分析 TCP BBR 在不同网络条件下的表现。这个示例对于 ns-3 新手来说是一个很好的入门案例,能够帮助他们理解 ns-3 的基本用法和网络仿真的基本概念。