[Ns-developers] Multirate 802.11 implementation is now available

[giuseppe de marco]

Federico Maguolo wrote:
> Hi all,
> I would like to announce the publication of a new 802.11 that we
> developed at the SIGNET lab, University of Padova, Department of
> Information Engineering (DEI), downloadable  at
>
> http://www.dei.unipd.it/wdyn/?IDsezione=5090
>
> This 802.11 implementation, named dei80211mr, permits the simulation of rate 
> adaptation algorithm (in the distribution an implementation of ARF  is already 
> included) with an enhanced physical layer with respect to those 
> in the ns distribution. In particular:
>
> - support for multiple PHY modes is included; in particolar,
>   dei80211mr supports simulation of the different transmission rates,
>   modulation and coding schemes defined in the IEEE802.11b/g standards.
>
> - a SINR-based packet level error model is introduced:
>   + the RX Threshold variable which was used in the 802.11
>     implementation included in standard NS to determine successful
>     receptions has been removed. Instead, Packet Error Rate (PER) is
>     used to determine random packet losses.
>   + PER is calculated using pre-determined curves (PER
>     vs SINR and packet size); the curves can be specified by the user
>     via TCL. Some default curves for both 802.11g and 802.11b are
>     provided.
>   + SINR is calculated using received signal strength, noise and
>     interference
>   + interference is calculated using a gaussian model to account for
>     all transmissions which happen simultaneously to the one which is
>     considered for reception
>   + noise power is set via TCL
>
> - the capture model, i.e. the determination of whether a packet can be
>   received when there are other concurrent transmissions are
>   simultaneously ogoing, is now embedded in the above mentioned
>   interference model (no more Capture Threshold).
>
> - Some well-known bugs of the 802.11 implementation in NS have been
>   resolved. With reference to [1], we solved the following bugs:
>   + direct access denial
>   + random backoff time
>   + capture model (note: this is different from the point above since,
>     as per the description in [1], it is actually a synchronization
>     issue)
>
> - In the wireless channel, the affected nodes distance is no more
>   determined using the CS threshold, but we used a fixed value in
>   meters (distInterference_) which can be set at the beginning of the
>   simulation. The reason is that, since we use a gaussian interference
>   model, nodes well below the CS threshold often still provide a
>   non-negligible contribution to interference. The default value for
>   the affected nodes distance (distInterference_) is very
>   conservative, so that all nodes are considered for interference
>   calculation. This default value therefore yields accurate but
>   computationally intensive simulations. The value can be adjusted via
>   TCL to achieve different trade-offs between computational load and
>   simulation accuracy.
>
> - Several useful 802.11 MAC MIB counters described in [3], Annex D, have 
>   been added to the MACMIB C++ class
>
> - dei80211mr can be used within the classical ns-2 MobileNode
>
>
>
> dei80211mr is distributed as a dynamic library using the Dynamic Library
> patch [2]. Thanks to the functionalities introduced by this patch,
> dei80211mr can be used with different versions of ns2. We have tested it
> with ns-2.29 and ns-2.31, and we expect it to work with future ns2
> releases as well.
>
>
> [1] Ilango Purushotaman and Sumit Roy,
> "IEEE802.11 implementation Issues in Network Simulator 2",
> Dept. of Electrical Engineering, University of Washington, US
> http://ee.washington.edu/research/funlab/
>
> [2] Patch for Loading Dynamic Modules in ns-2,
> http://mailman.isi.edu/pipermail/ns-users/2007-June/060486.html
> http://www.dei.unipd.it/~rossi/ns2-patch.html
>
> [3] IEEE LAN MAN Standards, Part 11: Wireless LAN
> Medium Access Control (MAC) and Physical Layer
> (PHY) specifications¿, ANSI/IEEE Std., March 1999.
>
>
>   
What do you mean with "gaussian model"?
I think the problem in simulation is not to use approximation but mimic 
what really happens.
Interference should be calculated as a cumulative interference, which 
varies along the reception
of a packet as the interferring packets arrives/depart.


-- 
Giuseppe De Marco, phD
Toyota Technological Institute
468-8511 Aichi 2-12-1 Hisakata, Tenpaku-ku, 
Nagoya, Japan
Email: demarco at toyota-ti dot ac dot jp
Tel (int): +81 (052)-809-1802
Skype-Id: giuseppe_dem2