Hi André,
I'm not sure I understood your question 100% but hoping this clarification might help. I haven't used LinkLayerModel myself, but looking at the documentation in the TinyOS code base (/doc/html/tutorial/usc-topologies.html) this appears to be a tool which attempts to solve a very difficult problem. The problem is: for a given scenario, how do we accurately state what the signal strengths will be between nodes in a network at any particular time during the simulation, given what we know about the target environment, without having to physically go out to the real environment, deploy some real nodes and physically measure a sample of the signal strengths. The reason this problem is so difficult is there are a huge number of complex and dynamic factors which will affect signal strengths in the real world - obviously static things like distance between nodes, but also dynamic things like how much RF interference there is in the air at that particular moment, atmospheric conditions, how signals bounce, refract and scatter off terrain and nearby objects, noise from external sources etc. Because of this complexity, any radio propagation model like TOSSIM's LinkLayerModel can only give you a best guess statistical model. Parts of this statistical model involve random variables which simulate some of the dynamic effects like RF noise, which are too complex and unknowable to predict absolutely. The documentation for LinkLayerModel describes a number of these random variables, for example: - Nodes are placed in random locations for the Random topology type, and in random places within a given cell for Uniform - Signal strength decay is random and log-normally distributed (the 'amount of randomness' is controlled by the SHADOWING_STANDARD_DEVIATION parameter) - Under 'Radio paratmeters' there is lots of discussion of dynamic variation in the model controlled by the standard deviation parameter WHITE_GAUSSIAN_NOISE for a Gaussian random process What I think this means is that if LinkLayerModel is supplied with exactly the same configuration file more than once, even though the configuration is the same, it will always generate a slightly different output in terms of the gain and noise over time for a simulation, because of the randomness in the model. So basically, I think what you are seeing is correct and intended behavior. Hope that helps. James ________________________________ From: tinyos-help-boun...@millennium.berkeley.edu <tinyos-help-boun...@millennium.berkeley.edu> on behalf of arsaraiva <andresara...@id.uff.br> Sent: 08 April 2017 13:04:58 To: tinyos-help@millennium.berkeley.edu Subject: [Tinyos-help] TOSSIM - Topology Hello friends, After staying until dawn yesterday trying to understand the differences, I decided to ask for help. I have 4 files: gera-topologia.txt, linkagain1.txt and linkagain2.txt. The file-topology file was taken from the TOSSIM tutorial itself and is exactly the same as that used by a teacher in 2010. TOSSIM has its own topology generator (LinkLayerModel) implemented in Java and in which it only works with the above file with correctly configured parameters. The file linkagain1.txt is the topology used by the teacher for 15 meters, according to his files in all 5 successful experiments. The file linkagain2.txt was created from the top-generation to the same 15 meters, and using the same generates topology. All the parameters are identical in the topology, but the linkagain's are very different. What is generating a very different behavior in my experiment, because if I use the teacher's file, it works fine, more if I use my, ZERO receive beacon or data. It is certainly due to the discrepancy between dB of the files, because already in the first lines you can see this: Linkagain1 Gain 0 1 -94.56 Gain 1 0 -94.80 Gain 0 2 -102.00 Linkagain2 Gain 0 1 -108.73 Gain 1 0 -109.45 Gain 0 2 -126.10 In http://tinyos-help.10906.n7.nabble.com/Re-Tossim-Topology-generation-td2334.html was posed the question of generation being doing high dB values. My initial question is this: Does my system with upgraded java (1.8), upgraded python (3.5), and the updated OS itself (Ubuntu 16) might be impacting on topology generation? An alternative that I found is to decrease in the gera-topology at a distance (in fact the values calculated by the teacher are close to something around 8 meters). Being that when increasing the distance to more than 10 meters, something strange happens, because it diminishes the reception of the beacons of synchronization and of the data packets (happened to only receive 2 beacons of synchronization), what is certain is the loss of packages, Because the sending confirms to me by SendDone that XXXX beacons have been sent. This happens with grid topology, random and random. Would anyone have a suggestion for this? Att. André -- View this message in context: http://tinyos-help.10906.n7.nabble.com/TOSSIM-Topology-tp25084.html Sent from the TinyOS - Help mailing list archive at Nabble.com. _______________________________________________ Tinyos-help mailing list Tinyos-help@millennium.berkeley.edu https://www.millennium.berkeley.edu/cgi-bin/mailman/listinfo/tinyos-help
_______________________________________________ Tinyos-help mailing list Tinyos-help@millennium.berkeley.edu https://www.millennium.berkeley.edu/cgi-bin/mailman/listinfo/tinyos-help
