Project report and presentation slides

My final project report can be found here and my presentation slides here.

Simulation of multihop relay scenario

I tried simulating a scenario consisting of multi hop relays (upto 3 hops). I referred to [1] for the algorithm for relay selection. You can find the pdf here and the matlab code here.

References:
[1]]S. Ann, K.G. Lee, and H.S. Kim "A Path Selection Method in IEEE 802.16j Mobile Multi-hop Relay Networks" International Conference on Sensor Technologies and Applications, Aug 2008.

Comparative Study of Cooperative Diversity Schemes

Last week I modelled some relaying schemes and evaluated their performance in a specific environmental setting. Today, I did a comparative study of different relaying schemes using the same system model. The pdf can be found here and the matlab code can be found here.

Next week I plan to simulate a relay selection algorithm as specified in [1]. I would also like to evaluate the loss of throughput we get with this algorithm when both the RS and MS are highly mobile.

The relay selection and routing in Wimax can either be done centrally at the BS or at the RS in a distributed fashion.

In a highly mobile environment distributed scheduling seems to be a better option. I will try to verify that.

Reference:
[1]S. Ann, K.G. Lee, and H.S. Kim "A Path Selection Method in IEEE 802.16j Mobile Multi-hop Relay Networks" International Conference on Sensor Technologies and Applications, Aug 2008.

Simulation Results

As of now, I tried out some basic simulations that show the improvement in throughput due to the use of relays. You can find the pdf here:

You can also find the matlab code here: norelay, simple_relay, coop_relay

Qualnet: a useful tool for modelling network scenarios?

This is in continuation of my previous post where I had a query regarding a suitable tool to model network scenarios. A quick check on google got me this one. Qualnet by Scalable Network Technologies seems to be useful tool. The website gives the following details:

"QualNet Developer is ultra high-fidelity network evaluation software that predicts wireless, wired and mixed-platform network and networking device performance. Designed to take full advantage of the multi-threading capabilities of multi-core 64-bit processors, QualNet supports simulation of thousands of network nodes. QualNet Architect Design Mode allows users to set up terrain, network connections, subnets, mobility patterns of wireless users, and other functional parameters of network nodes. Users can create network models by using intuitive, click and drag operations. Users can also customize the protocol stack of any of the nodes and can specify the application layer traffic and services that run on the network"


I think it is a C++ based tool and you need to spend some time reading the user manual and some APIs to model your own project. But I think it is worth a try. My only concern is getting a licensed copy. I hope it is available somewhere in the department. Students are eligible to get 30 days free trial version. But in that case you have to wait till someone from the company contacts you after you register on their website. I am still waiting :)

Cooperative diversity for Wimax: Practical schemes for scheduling and scheme selection

I found this nice paper that proposes methods and ways to deal with cooperative diversity scheme selection and radio resource allocation that could be used in the 802.16j standard which is still under development.
http://www.sce.carleton.ca/~furuzan/files/papers/wcnc08-bc.pdf

Since it is high time now to decide the exact nature of the project, I realized that it is also important to have a presentable model and some simulation results apart from just gaining knowledge by reading theory. This paper provides a useful guiding example for designing my own model.

Some brief overview of this paper is as follows:

1. Considers two hop network and uses OFDMA PHY for multiple access.
2. Relays are decode and forward type.
3. All cooperative diversity schemes use only one relay to realize diversity
4. Relaying schemes could be one of the following:
i) No relays (direct transmission)
ii) Relay used without diversity (just to improve coverage)
iii) Relay used with cooperative diversity (Several different flavors of this are used: refer to the paper for more details)
iv) Adaptive diversity (combination of the above)


The choice of a particular diversity scheme depends on the calculations done using the instantaneous SNRs in the BS to MS and RS to MS channels. A look up table helps to speed up the calculations. It also considers opportunistic scheduling of the users based on instantaneous SNRs.

The system model assumes practical modulation schemes, subchannel allocations, symbol times, coherence times, etc (details skipped here), Rayleigh fading is considered for all the channels and it is flat fading due to OFDMA.

The simulation results are plotted as the comparative gains of different diversity schemes in directional graph for a single cell with single base. An example of such a graph is shown above.

My first task will be modelling a similar system using suitable tool (Qualnet?). I am still not sure as to what details I should look at. However a simplistic model should be a good starting point. I would like to also evaluate the performance of diversity schemes with more than one relay and combine the process fo relay selection with all of this so that we get the maximum throughput.

Simulating relay selection algorithms

Although I am not quite ready to simulate anything substantial for this project I had few questions and concerns.

I was just wondering if there is some easy way way of simulating a relay network? Matlab is what came to my mind foremost. Does anyone know of a mathematical model of a network channel that can be just picked up and used for simulation?

Also, if I just want to study the system throughput of different relay selection algorithms, how precise should my trnsreceiver block in Matlab be? By precise I mean should I include the errors due to frequency offsets? Should I include the convolution encoder/decoder?

Your help will be much appreciated :)