Extend pathLossMap.txt with the existing values

[Dimitra Kaitalidou]

Dear all,
I am trying to simulate a body area network with more than 6 nodes and I couldn't find a data set with pathloss values that fit the simulator (this
http://nrg.nicta.com.au/people/current/david-smith/ didn't help me).
Therefore I decided to increase the number by placing more than one nodes on the same body parts as the initial simulation model.
For example if I want to simulate 30 nodes instead of just 6 I extended the pathLossMap file as follows:

#0, 10, 20 are R-hip
#1, 6, 11, 16, 21, 26 are L-wrist
#2, 7, 12, 17, 22, 27 are R-wrist
#3, 8, 13, 18, 23, 28 are L-ankle
#4, 9, 14, 19, 24, 29 are R-ankle
#5, 10, 15, 25 are chest

In the pathLossMap.txt file it stated that: 0>1:56 and therefore I added 10>11:56, 20>21:56, 10>21 :56, 20>11:56 and so on, since the nodes 10, 20 and 11, 21 are placed where the nodes 0 and 1 are places respectively.

I have simulated 10, 20, 30, 40 and 50 nodes and for each simulation there is a dedicated (extended as described above) pathloss file, which is saved under the directory of the original pathLossMap.txt file and is included in the omnetpp.ini before each simulation. The results of the simulations seem normal, but I am still not sure.

What do you think of this choice? Is it acceptable?
Would it be better to use the lognormal shadowing model?

Best regards!

[Hashin Jithu]

The path loss model in Castalia was developed with a man running on a treadmill with these 6 sensor nodes attached to him. Also, when you write 10>11:56, you are meaning that when node 10 sends data to node 11, it experiences a path loss of 56 dBm. So the values given may not be suitable for your application with a lot of nodes.

Also, lognormal shadowing is not a great approximation for the Body area networks. I would suggest that you try to develop a path loss model if possible. Or if you can ensure that the interference caused by your devices are small, you may carry on with your current approach.

Hope it helps.

[Dimitra Kaitalidou]

Thank you very much for your response. Reading the Castalia User Manual I couldn't find information about the movement of the man from whom the pathloss measurements were taken. Could you clarify where lies this information ("a man running on a treadmill ")?

[Hashin Jithu]

That was stated by Prof. Athanassios under another thread in this group. You can find it here. Hope that it helps. 

--
You received this message because you are subscribed to a topic in the Google Groups "Castalia Simulator" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/castalia-simulator/B9YfJRhNkzQ/unsubscribe.
To unsubscribe from this group and all its topics, send an email to castalia-simula...@googlegroups.com.
To post to this group, send email to castalia-...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/castalia-simulator/69babca3-2e00-4cc5-906b-53ba040b76f7%40googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

[Thanassis Boulis]

On 29 November 2014 at 02:39, Dimitra Kaitalidou <dkai...@gmail.com> wrote:

Dear all,
I am trying to simulate a body area network with more than 6 nodes and I couldn't find a data set with pathloss values that fit the simulator (this
http://nrg.nicta.com.au/people/current/david-smith/ didn't help me).

David does not provide files that are ready to be used with Castalia, but rather provides a lot of raw measurements, that can help you infer/calculate/create Castalia-compatible pathloss maps. I would start with this information.

 

Therefore I decided to increase the number by placing more than one nodes on the same body parts as the initial simulation model.
For example if I want to simulate 30 nodes instead of just 6 I extended the pathLossMap file as follows:

#0, 10, 20 are R-hip
#1, 6, 11, 16, 21, 26 are L-wrist
#2, 7, 12, 17, 22, 27 are R-wrist
#3, 8, 13, 18, 23, 28 are L-ankle
#4, 9, 14, 19, 24, 29 are R-ankle
#5, 10, 15, 25 are chest

You understand that the above are just comments, and do not really extend anything. Maybe you just meant them as a description of what you wanted to do.

 

In the pathLossMap.txt file it stated that: 0>1:56 and therefore I added 10>11:56, 20>21:56, 10>21 :56, 20>11:56 and so on, since the nodes 10, 20 and 11, 21 are placed where the nodes 0 and 1 are places respectively.

Not sure what "so on" means.  Have you created a pathloss map with 30x30 = 900 values?  

You noticed how the original map has 36 values for all 6x6 possible path combinations. If you need to extend the pathloss to 30 nodes where groups of 5 nodes are bundled in one spot, then you can use these 36 original values, and copy them in approprate places of this 30x30 combinations pathloss map

If you do not include a combination, I think it is taken as a pathloss of 0 (not sure, you'll have to look it up)

Not sure how useful a scenario like this would be. I would definitely urge you to use David's data to produce a more realistic pathloss map.

[Dimitra Kaitalidou]

Thank you very much for your response Dr. Boulis.

In the pathLossMap.txt file it stated that: 0>1:56 and therefore I added 10>11:56, 20>21:56, 10>21 :56, 20>11:56 and so on, since the nodes 10, 20 and 11, 21 are placed where the nodes 0 and 1 are places respectively.

Not sure what "so on" means.  Have you created a pathloss map with 30x30 = 900 values?  

You noticed how the original map has 36 values for all 6x6 possible path combinations. If you need to extend the pathloss to 30 nodes where groups of 5 nodes are bundled in one spot, then you can use these 36 original values, and copy them in approprate places of this 30x30 combinations pathloss map.

 I did create a 30x30 matrix of pathloss measurements (900 values in total) exactly as you described.

If you do not include a combination, I think it is taken as a pathloss of 0 (not sure, you'll have to look it up)

By declaring N number of nodes and no pathloss map file or no location for each node, one does use pathloss values of 0, since a node's location is by default (0,0,0) on the simulator's coordinate system. Therefore all N nodes have 0 distance from each other and, according to the pathloss equation in WirelessChannel.cc, 0 pathloss between them.

Knowing this one has two options:
1. Declare N nodes with the respective location of each and use (by default) the lognormal shadowing model implemented in WirelessChannel.cc file.
2. Use a pathloss map file with NxN values and skip the model mentioned in (1).

Since it is stated in the User manual that the lognormal shadowing model does not produce good results for BAN I should go with the second option.

David does not provide files that are ready to be used with Castalia, but rather provides a lot of raw measurements, that can help you infer/calculate/create Castalia-compatible pathloss maps. I would start with this information.

 

I have studied the dataset you propose, but for increased number of nodes (the maximum number of nodes used in these measurements is 15) the carrier frequency used is 820MHz. I am only interested in 2.4GHz and for this carrier frequency there are no measurements for a number of nodes larger than 6.

Best Regards!
 

[Thanassis Boulis]

By declaring N number of nodes and no pathloss map file or no location for each node, one does use pathloss values of 0, since a node's location is by default (0,0,0) on the simulator's coordinate system. Therefore all N nodes have 0 distance from each other and, according to the pathloss equation in WirelessChannel.cc, 0 pathloss between them.

Ah right. And I know remembered that if you declare the positions of two nodes but not define a pathloss between them, then the default lognormal shadowing model will be used.

Knowing this one has two options:
1. Declare N nodes with the respective location of each and use (by default) the lognormal shadowing model implemented in WirelessChannel.cc file.
2. Use a pathloss map file with NxN values and skip the model mentioned in (1).

There is a 3rd option, a hybrid. Some pairs have the pathloss defined in the map, and the ones that don't get a pathloss from the lognormal shadowing model.

 

Since it is stated in the User manual that the lognormal shadowing model does not produce good results for BAN I should go with the second option.

 

Yes.

 

David does not provide files that are ready to be used with Castalia, but rather provides a lot of raw measurements, that can help you infer/calculate/create Castalia-compatible pathloss maps. I would start with this information.

 

I have studied the dataset you propose, but for increased number of nodes (the maximum number of nodes used in these measurements is 15) the carrier frequency used is 820MHz. I am only interested in 2.4GHz and for this carrier frequency there are no measurements for a number of nodes larger than 6.

Yes, you will not find more than 15 node positions in David's datasets. Not sure how the frequency affects results. You might find David's papers that discuss this. I think the frequency response is pretty flat (I might be mistaken though), so it would be a good approximation to use david's datasets. Of course the ideal situation would be to run your own measurements/experiments, but I understand this is quite difficult and expensive in both time and cost.   

 

[marwen amiri]

Hello Dear Boulis, 

Hello Dear Castalia Users

i hope you are doing well, Actually, I am working on the modeling of path loss between the coordinator and the sensor nodes of a BAN network. My objective is to make a performance comparison between the CM3A model of the IEEE802.15.6 standard and a loss model that I have implemented mathematically.

 I created two pathloss Map files the first for CM3 and the second is for my model. My problem is that every time I run the simulation, Castalia runs the simulations for CM3 pathlossMAP file (IEEE 802.15.6) but it does not run the simulations for my pathlossMap file (AlphaBody model)

Can you give me according to your respectful experience an explanation for this problem? Why CASTALIA does not accept the path loss values added in the alphaBody.txt file?

Thanks in advance

i attach the two pathloss Maps (CM3 and alpha Body)

thanks

[marwen amiri]

Hello Dear Castalia Users

Hello Mr Boulis, 

i hope you are doing well, Actually, I am working on the modeling of path loss between the coordinator and the sensor nodes of a BAN network. My objective is to make a performance comparison between the CM3A model of the IEEE802.15.6 standard and a loss model that I have implemented mathematically.

 I created two pathloss Map files the first for CM3 and the second is for my model. My problem is that every time I run the simulation, Castalia runs the simulations for CM3 pathlossMAP file (IEEE 802.15.6) but it does not run the simulations for my pathlossMap file (AlphaBody model)

Can you give me according to your respectful experience an explanation for this problem? Why CASTALIA does not accept the path loss values added in the alphaBody.txt file?

Thanks in advance

i attach the two pathloss Maps (CM3 and alpha Body)

thanks