@node Emulation

@chapter Emulation

@anchor{chap:Emulation}

ns-3 has been designed for integration into testbed and virtual machine

environments.  We have addressed this need by providing two kinds of 

net devices.  The first kind, which we call an @code{Emu} @code{NetDevice}

allows ns-3 simulations to send data on a ``real'' network.  The second kind,

called a @code{Tap} @code{NetDevice} allows a ``real'' host to participate 

in an ns-3 simulation as if it were one of the simulated nodes.  An ns-3 

simulation may be constructed with any combination of simulated, @code{Emu}, 

or @code{Tap} devices.

One of the use-cases we want to support is that of a testbed.  A concrete 

example of an environment of this kind is the ORBIT testbed.  ORBIT is a 

laboratory emulator/field trial network arranged as a two dimensional grid of

400 802.11 radio nodes.  We integrate with ORBIT by using their ``imaging'' 

process to load and run ns-3 simulations on the ORBIT array.  We use our

@code{Emu} @code{NetDevice}s to drive the hardware in the testbed and we can

accumulate results either using the ns-3 tracing and logging functions, or the 

native ORBIT data gathering techniques.  See @uref{http://www.orbit-lab.org/} 

for details on the ORBIT testbed.

A simulation of this kind is shown in the following figure:

@float Figure,fig:testbed

@center @caption{Example Implementation of Testbed Emulation.}

@center @image{figures/testbed, 5in}

@end float

You can see that there are separate hosts, each running a subset of a ``global''

simulation.  Instead of an ns-3 channel connecting the hosts, we use real

hardware provided by the testbed.  This allows ns-3 applications and protocol

stacks attached to a simulation node to communicate over real hardware.

We expect the primary use for this configuration will be to generate repeatable

experimental results in a real-world network environment that includes all of 

In what can be viewed as essentially an inverse configuration, we allow ``real'' 

machines running native applications and protocol stacks to integrate with

an ns-3 simulation.  This allows for the simulation of large networks connected

to a real mahince, and also enables virtualization.  A simulation of this kind

is shown in the following figure:

@float Figure,fig:emulated-channel

@caption{Implementation overview of emulated channel.}

@image{figures/emulated-channel, 5in}

@end float

Here, you will see that there is a single host with a number of virtual machines 

running on it.  An ns-3 simulation is shown running in the virtual machine shown

in the center of the figure.  This simulation has a number of nodes with associated 

ns-3 applications and protocol stacks that are talking to an ns-3 channel through 

native simulated ns-3 net devices.

There are also two virtual machines shown at the far left and far right of the 

figure.  These VMs are running native (Linux) applications and protocol stacks.

The VM is connected into the simulation by a Linux Tap net device.  The user-mode

handler for the Tap device is instantiated in the simulation and attached to a 

proxy node that represents the native VM in the simulation.  These handlers allow

the Tap devices on the native VMs to behave as if they were ns-3 net devices in 

the simulation VM.  This, in turn, allows the native software and protocol suites

in the native VMs to believe that they are connected to the simulated ns-3

channel.

We expect the typical use case for this environment will be to analyze the 

behavior of native applications and protocol suites in the presence of large 

simulated ns-3 networks.

@section Behavior

@subsection Emu Net Device

The @code{Emu} net device allows a simulation node to send and receive packets

over a real network.  The emulated net device relies on a specified interface 

being in promiscuous mode.  It opens a raw socket and binds to that interface.

We perform MAC spoofing to separate simulation network traffic from other 

network traffic that may be flowing to and from the host.

Normally, the use case for emulated net devices is in collections of small 

simulations that connect to the outside world through specific interfaces.

For example, one could construct a number of virtual machines and connect them

via a host-only network.  To use the emulated net device, you would need to 

set all of the host-only interfaces in promiscuous mode and provide an 

appropriate device name, "eth1" for example.

One could also use the @code{Emu} net device in a testbed situation where the 

host on which the simulation is running has a specific interface of interest 

which  drives the testbed hardware.  You would also need to set this specific 

interface into promiscuous mode and provide an appropriate device name to the 

ns-3 emulated net device.  An example of this environment is the ORBIT testbed 

as described above.

The @code{Emu} net device only works if the underlying interface is up and in 

promiscuous mode.  Packets will be sent out over the device, but we use MAC

spoofing.  The MAC addresses will be generated (by default) using the 

Organizationally Unique Identifier (OUI) 00:00:00 as a base.  This vendor code

is not assigned to any organization and so should not conflict with any real 

hardware.  

It is always up to the user to determine that using these MAC addresses is

okay on your network and won't conflict with anything else (including another

simulation using @code{Emu} devices) on your network.  If you are using the 

emulated net device in separate simulations you must consider global MAC 

address assignment issues and ensure that MAC addresses are unique across

all simulations.  The emulated net device respects the MAC address provided

in the @code{SetAddress} method so you can do this manually.  For larger 

simulations, you may want to set the OUI in the MAC address allocation function.

IP addresses corresponding to the emulated net devices are the addresses 

generated in the simulation, which are generated in the usual way via helper

functions.  Since we are using MAC spoofing, there will not be a conflict 

between ns-3 network stacks and any native network stacks.

The emulated net device comes with a helper function as all ns-3 devices do.

One unique aspect is that there is no channel associated with the underlying

medium.  We really have no idea what this external medium is, and so have not

made an effort to model it abstractly.  The primary thing to be aware of is the 

implication this has for static global routing.  The global router module

attempts to walk the channels looking for adjacent networks.  Since there 

is no channel, the global router will be unable to do this and you must then 

use a dynamic routing protocol such as OLSR to include routing in 

@code{Emu}-based networks.

@subsection Tap Net Device

The @code{Tap} Net Device is scheduled for inclusion in ns-3.4 at the writing

of this section.  We will include details as soon as the @code{Tap} device is

merged.

@section Usage

Any mixing of ns-3 objects with real objects will typically require that

ns-3 compute checksums in its protocols.  By default, checksums are not

computed by ns-3.  To enable checksums (e.g. UDP, TCP, IP), users must set

the attribute @code{ChecksumEnabled} to true, such as follows:

@verbatim

GlobalValue::Bind ("ChecksumEnabled", BooleanValue (true));

@end verbatim

@subsection Emu Net Device

The usage of the @code{Emu} net device is straightforward once the network of

simulations has been configured.  Since most of the work involved in working 

with this device is in network configuration before even starting a simulation,

you may want to take a moment to review a couple of HOWTO pages on the ns-3 wiki

that describe how to set up a virtual test network using VMware and how to run

a set of example (client server) simulations that use @code{Emu} net devices.

@uref{http://www.nsnam.org/wiki/index.php/HOWTO_use_VMware_to_set_up_virtual_networks_(Windows)}

@uref{http://www.nsnam.org/wiki/index.php/HOWTO_use_ns-3_scripts_to_drive_real_hardware_(experimental)} 

Once you are over the configuration hurdle, the script changes required to use 

an @code{Emu} device are trivial.  The main structural difference is that you

will need to create an ns-3 simulation script for each node.  In the case of

the HOWTOs above, there is one client script and one server script.  The only

``challenge'' is to get the addresses set correctly.

Just as with all other ns-3 net devices, we provide a helper class for the 

@code{Emu} net device.  The following code snippet illustrates how one would

declare an EmuHelper and use it to set the ``DeviceName'' attribute to ``eth1''

and install @code{Emu} devices on a group of nodes.  You would do this on both

the client and server side in the case of the HOWTO seen above.

@verbatim

  EmuHelper emu;

  emu.SetAttribute ("DeviceName", StringValue ("eth1"));

  NetDeviceContainer d = emu.Install (n);

@end verbatim

The only other change that may be required is to make sure that the address

spaces (MAC and IP) on the client and server simulations are compatible.  First

the MAC address is set to a unique well-known value in both places (illustrated

here for one side).

@verbatim

  //

  // We've got the devices in place.  Since we're using MAC address 

  // spoofing under the sheets, we need to make sure that the MAC addresses

  // we have assigned to our devices are unique.  Ns-3 will happily

  // automatically assign the same MAC addresses to the devices in both halves

  // of our two-script pair, so let's go ahead and just manually change them

  // to something we ensure is unique.

  //

  Ptr<NetDevice> nd = d.Get (0);

  Ptr<EmuNetDevice> ed = nd->GetObject<EmuNetDevice> ();

  ed->SetAddress ("00:00:00:00:00:02");

@end verbatim

helpers.

@verbatim

  //

  // We've got the "hardware" in place.  Now we need to add IP addresses.

  // This is the server half of a two-script pair.  We need to make sure

  // that the addressing in both of these applications is consistent, so

  // we use provide an initial address in both cases.  Here, the client 

  // will reside on one machine running ns-3 with one node having ns-3

  // with IP address "10.1.1.2" and talk to a server script running in 

  // another ns-3 on another computer that has an ns-3 node with IP 

  // address "10.1.1.3"

  //

  Ipv4AddressHelper ipv4;

  ipv4.SetBase ("10.1.1.0", "255.255.255.0", "0.0.0.2");

  Ipv4InterfaceContainer i = ipv4.Assign (d);

@end verbatim

You will use application helpers to generate traffic exactly as you do in any

ns-3 simulation script.  Note that the server address shown below in a snippet

from the client, must correspond to the IP address assigned to the server node

similarly to the snippet above. 

@verbatim

  uint32_t packetSize = 1024;

  uint32_t maxPacketCount = 2000;

  Time interPacketInterval = Seconds (0.001);

  UdpEchoClientHelper client ("10.1.1.3", 9);

  client.SetAttribute ("MaxPackets", UintegerValue (maxPacketCount));

  client.SetAttribute ("Interval", TimeValue (interPacketInterval));

  client.SetAttribute ("PacketSize", UintegerValue (packetSize));

  ApplicationContainer apps = client.Install (n.Get (0));

  apps.Start (Seconds (1.0));

  apps.Stop (Seconds (2.0));

@end verbatim

The @code{Emu} net device and helper provide access to ASCII and pcap tracing

functionality just as other ns-3 net devices to.  You enable tracing similarly

to these other net devices:

@verbatim

  EmuHelper::EnablePcapAll ("emu-udp-echo-client");

@end verbatim

To see an example of a client script using the @code{Emu} net device, see

@code{examples/emu-udp-echo-client.cc} and @code{examples/emu-udp-echo-server.cc}

in the repository @uref{http://code.nsnam.org/craigdo/ns-3-emu/}. 

@subsection Tap Net Device

The @code{Tap} Net Device is scheduled for inclusion in ns-3.4 at the writing

of this section.  We will include details as soon as the @code{Tap} device is

merged.

@section Implementation

Perhaps the most unusual part of the @code{Emu} and @code{Tap} device 

implementation relates to the requirement for executing some of the code 

with super-user permissions.  Rather than force the user to execute the entire

simulation as root, we provide a small ``creator'' program that runs as root

and does any required high-permission sockets work.

We do a similar thing for both the @code{Emu} and the @code{Tap} devices.

The high-level view is that the @code{CreateSocket} method creates a local 

interprocess (Unix) socket, forks, and executes the small creation program.

The small program, which runs as suid root, creates a raw socket and sends 

back the raw socket file descriptor over the Unix socket that is passed to

it as a parameter.  The raw socket is passed as a control message (sometimes 

called ancillary data) of type SCM_RIGHTS.

@subsection Emu Net Device

The @code{Emu} net device uses the ns-3 threading and multithreaded real-time

scheduler extensions.  The interesting work in the @code{Emu} device is done

when the net device is started (@code{EmuNetDevice::StartDevice ()}).  An 

attribute (``Start'') provides a simulation time at which to spin up the 

net device.  At this specified time (which defaults to t=0), the socket 

creation function is called and executes as described above.  You may also

specify a time at which to stop the device using the ``Stop'' attribute.

Once the (promiscuous mode) socket is created, we bind it to an interface name 

also provided as an attribute (``DeviceName'') that is stored internally as 

@code{m_deviceName}:

@verbatim

  struct ifreq ifr;

  bzero (&ifr, sizeof(ifr));

  strncpy ((char *)ifr.ifr_name, m_deviceName.c_str (), IFNAMSIZ);

  int32_t rc = ioctl (m_sock, SIOCGIFINDEX, &ifr);

  struct sockaddr_ll ll;

  bzero (&ll, sizeof(ll));

  ll.sll_family = AF_PACKET;

  ll.sll_ifindex = m_sll_ifindex;

  ll.sll_protocol = htons(ETH_P_ALL);

  rc = bind (m_sock, (struct sockaddr *)&ll, sizeof (ll));

@end verbatim

After the promiscuous raw socket is set up, a separate thread is spawned to do 

reads from that socket and the link state is set to @code{Up}.

@verbatim

  m_readThread = Create<SystemThread> (

    MakeCallback (&EmuNetDevice::ReadThread, this));

  m_readThread->Start ();

  NotifyLinkUp ();

@end verbatim

The @code{EmuNetDevice::ReadThread} function basically just sits in an infinite

loop reading from the promiscuous mode raw socket and scheduling packet 

receptions using the real-time simulator extensions.

@verbatim

  for (;;)

    {

      ...

      len = recvfrom (m_sock, buf, bufferSize, 0, (struct sockaddr *)&addr, 

        &addrSize);

      ...

      DynamicCast<RealtimeSimulatorImpl> (Simulator::GetImplementation ())->

        ScheduleRealtimeNow (

          MakeEvent (&EmuNetDevice::ForwardUp, this, buf, len));

      ...

    }

@end verbatim

The line starting with our templated DynamicCast function probably deserves a 

comment.  It gains access to the simulator implementation object using

the @code{Simulator::GetImplementation} method and then casts to the real-time

simulator implementation to use the real-time schedule method 

@code{ScheduleRealtimeNow}.  This function will cause a handler for the  newly

received packet to be scheduled for execution at the current real time clock 

value.  This will, in turn cause the simulation clock to be advanced to that 

real time value when the scheduled event (@code{EmuNetDevice::ForwardUp}) is

fired.

The @code{ForwardUp} function operates as most other similar ns-3 net device 

methods do.  The packet is first filtered based on the destination address.  In 

the case of the @code{Emu} device, the MAC destination address will be the 

address of the @code{Emu} device and not the hardware address of the real 

device.  Headers are then stripped off and the trace hooks are hit.  Finally,

the packet is passed up the ns-3 protocol stack using the receive callback 

function of the net device.

Sending a packet is equally straightforward as shown below.  The first thing

we do is to add the ethernet header and trailer to the ns-3 @code{Packet} we

are sending.  The source address corresponds to the address of the @code{Emu}

device and not the underlying native device MAC address.  This is where the

MAC address spoofing is done.  The trailer is added and we enqueue and dequeue

the packet from the net device queue to hit the trace hooks.

@verbatim

  header.SetSource (source);

  header.SetDestination (destination);

  header.SetLengthType (packet->GetSize ());

  packet->AddHeader (header);

  EthernetTrailer trailer;

  trailer.CalcFcs (packet);

  packet->AddTrailer (trailer);

  m_queue->Enqueue (packet);

  packet = m_queue->Dequeue ();

  struct sockaddr_ll ll;

  bzero (&ll, sizeof (ll));

  ll.sll_family = AF_PACKET;

  ll.sll_ifindex = m_sll_ifindex;

  ll.sll_protocol = htons(ETH_P_ALL);

  rc = sendto (m_sock, packet->PeekData (), packet->GetSize (), 0, 

    reinterpret_cast<struct sockaddr *> (&ll), sizeof (ll));

@end verbatim

Finally, we simply send the packet to the raw socket which puts it out on the 

real network.

@subsection Tap Net Device

The @code{Tap} Net Device is scheduled for inclusion in ns-3.4 at the writing

of this section.  We will include details as soon as the @code{Tap} device is

merged.