Hi Bastian,
This looks really useful and promising, thanks!
However, in general, I would avoid to setup large scale ns-3 simulations that model the phy on signal level. That get’s complicated and computationally complex. Maybe it is
possible to make phy simulations, derive a simulation model from the results, and use the model in ns-3.
I’m assuming you mean “large scale” in the number of devices. I’m only looking to simulate PHY + MAC for ~10-20 devices.
I am most interested in some random variable distributions related to MAC timing, such as SIFS. I know you had a WiNTECH paper which gives some good insight, but looks like you logged it off of the WLAN card. I’m thinking for field data, just
modify gr-IEEE 802.11g to measure this. With a USRP + GPSDO, timestamps should be quite good. For simulation, I’ll try PhySimWIFI.
PWG
Hi,
On 2 Mar 2017, at 16:27, Garver, Paul W <address@hidden> wrote:
Hey all,
I’m looking at doing some IEEE 802.11g simulations for my research and I’d like to solicit your opinion as to the best approach. Essentially, I think I have a method of improving localization accuracy combining TDoA with some information from the MAC Layer.
I’d like to simulate my idea in more fidelity than MATLAB. By this, I mean creating nodes which actually implement at least the PHY and MAC OSI Layers of the stack. gr-IEEE802.11 [1] has some nice support for the PHY layer, but critical MAC layer features
such as CSMA/CA are not implemented. I also want to simulate particular frame exchange sequences such as RTS/CTS. NS2/NS3 [2] appears to have support for many of the higher layer functions (e.g. MAC), but it’s unclear to me how much you can play around and
access the PHY Layer. I’m not very familiar with NS/NS3 in general. I need very precise (sub-sample level) information on packet arrival times, etc.
I’m wondering if I could combine the accessibility of the PHY layer in GNURadio with the MAC-Layer implementations in NS2/NS3. Basically, the idea is to let GR be the “modem” for NS2/NS3, and then try and combine the NS2/NS3 results with some output from the
GR simulations (e.g. time of arrival of particular packets) to get what I want.
Has anyone tried this? I appreciate your advice regarding the easiest path here...
sounds
like your looking for something like PhySimWiFi [1]. There is a good invited paper about it and Jens Mittags dissertation.
However,
in general, I would avoid to setup large scale ns-3 simulations that model the phy on signal level. That get’s complicated and computationally complex. Maybe it is possible to make phy simulations, derive a simulation model from the results, and use the model
in ns-3.
Best,
Bastian
[1] https://dsn.tm.kit.edu/english/ns3-physim.php
|