[Ns-developers] GSOC 2010: Cognitive Radio Networks Proposal

Tue Apr 6 07:21:49 PDT 2010

On 06/04/2010, at 06:53, Omar Jan wrote:

> Dear Andre,
> 
> Let me brief myself, I'm Ph.D student in TU-Ilmenau/Germany. Currently I'm
> working in CRN also.
> 
> Your suggestion is really interesting. It would helps many researchers.
> Basically I'm working in NS2 and due to NS2 does not support CRN such as
> multi interface or PU transmission. I'm working to develop these aspect. My
> research is on routing protocol to sum up you the routing algorithm should
> be able to:
> 1. select the Radio / Channel according to some information (PU statistics,
> Channel characteristics ... etc)
> 2. discovery the best route according to the spectrum information. 
> 3. apply different recovery strategies.
> 
> I would be glad to work in NS3 also. So please find my comments below. 
> 

Hi Omar! First of all, I really appreciate your comments.

> The following are the
>> main implementation and definitions suggestions for this proposal:
>> 
>> 1- A module that can simulate different PU transmissions based on
>> configurable statistical distributions of ON/OFF states and durations.
> 
> [Omar] As I understood, it should be two mobile node object, one for PU and
> other for SU. The PU object must accept realistic statistical information.
> Based on my knowledge it's implementable. 

>From what I've studied, the PUs can be simulated by an OnOff  Application. NS-3 already has such an application, but I consider it should be adapted to CNR to better represent real world conditions. The SUs would be nodes, yes.

> 
>> 2- Identification and definition of important metrics that can be used
>> to compare different models in Cognitive Radio Networks.
> 
> [Omar] The metrics based on several criteria which one of them is the PU
> statistic 

I have already collected some metrics that I find important, but I would like to discuss them with you or anyone interested to reach a common point.

> 
>> 3- A configurable Base Station module, which could accept different
>> spectrum decision algorithms and routing protocols.
> 
> [Omar] you mean it should be a comprehensive way that can be flexible way to
> modify. 

Yes, that's correct. I'm thinking about a way for researchers to implement their sensing or routing algorithms and through an interface, connect them to the Base Station.

> 
>> 4- A node object that could be configured to work in Ad Hoc mode or
>> Infrastructured mode. In both modes, it could accept different
>> implementations of spectrum decision protocols, sensing protocols and
>> MAC protocols and mobility algorithms. In Ad Hoc mode, it could also
>> have different routing protocols to handle traffic between nodes and
>> possibly an option for collaborative sensing.
> 
> [Omar] it would be so a little complicated as based on my knowledge but it
> is implementable. The first thing it must specify the applicable protocols
> in spectrum decision, sensing ... etc.
> 

This nodes would be abstractions for SUs. I should put more thought on this one because they can get really complex indeed. Maybe two node implementations would be needed, a CR node and a subclass of the latter, making a CR AdHoc node.....

>> 5- Identify the most used MAC protocols in Cognitive Radio literature
>> and implement at least one of those.
> 
> [Omar] this is important issue since the routing protocol must know about
> the MAC protocol strategy, I would proposed to indentify the routing
> protocols also in this part.

You are right, that's good advice.

> 
>> All such configurable modules or objects compose a general scenario of
>> a Cognitive Radio Network. As there is no defined standard, researchers
>> should be able to simulate diverse approaches to all main areas in
>> Cognitive Radio devices.
>> 
>> These suggestions should be all implemented in a best case scenario,
>> but any subset of them would give a good start into achieving a
>> trustworthy simulation environment.
> 
> [Omar] I'm working to implement the multi interfaces with PU transmission
> right now. I do not know if it's usefulness for you or not. But I'm so
> interesting to work with you in this project.

Gladly, NS-3 already has support for multi interfaces nodes, but in the case of CR, we need an interface that can support various channels.

My propositions involve a lot of work and I'll approach just part of that in my GSOC project, there will be a lot to do. It would be great to work with you too and besides, you're probably a more experienced researcher and help is surely wanted :)

>> 
> 
> 
> ---
> 
> Omar Jan
> 
> Integrated Communication Systems Group
> Institute of Computer Engineering
> Faculty of Computer Science and Automation
> Ilmenau University of Technology
> 
> Tel   : +49 3677 69 1203 
> Fax   : +49 3677 69 1614
> Email : omar-hussein.jan at tu-ilmenau.de
> 
> 
> 
> 

Andre Lauar Sampaio Meirelles
--------------------------------------------
Masters Student - Federal University of Goiás
Brazil
andrelauar at inf.ufg.br