[Draft] Basic Tracefile with Dynamic Vehicle Entry/Exit + some other fixes #37

chestnutcase · 2024-04-22T07:10:35Z

Hi V2XSim Maintainers,

I have a working version of the sim with dynamic vehicle entry/exit partially supported for you to review. The approach/constraints are as follows:

The maximum number of active vehicles throughout the simulation is set constant at the beginning of the simulation.
The vehicle ids in the tracefile must be consecutive integers from 1 to XXX (use a python script to preprocess the bologna tracefiles)
A rectangle bound in world coordinates is specified using XminTrace, XmaxTrace, YminTrace, YmaxTrace
To mark a vehicle as active, the tracefile has a log line that shows the vehicle within these bounds
When a vehicle exits from the simulation, the tracefile adds a log line in the next epoch that "teleports" the vehicle outside these bounds
Vehicles "inactive" stay in some arbitary place outside of the active rectangle bound and do not request BR etc. They still tick for position updates and physics calculations (change in LOS/shadowing), but do not schedule any packets so the centralized algorithms do not schedule them any BR.
For the vehicle to become active again, the tracefile adds another log line to teleport the vehicle back within these bounds

Using this approach I have a rudimentary support of dynamic vehicle entry/exit. The ideal scenario would be to exclude inactive vehicles from physics calculations entirely to speed up the simulation, but currently it is good enough for my current studies that requires tracefiles.

My studies is specifically 5GNR centralised mode, so I will need your help to review the PR to test the other modes (decentralised, and maybe vehicular wifi)

This PR contains some other files i accidentally committed from many months ago like MatFilesVisualization for my own studies but not fully functional by themselves yet. But the bulk of the changes that supports the tracefile is in the last commit.

Attached is the modified BolognaAPositions with the constraints (consecutive vehicle numbers, and "teleporting" inactive vehicles to coordinates 0,0 ), preprocessed using Python.
BolognaAPositions.csv

…n to manually specify simID

1. When writing the "mainout" file, using a "marker file" to indicate that the "mainout" file is writing now. In case of multiple processes write the file simultaneously. 2. If the outputs of different simulations are in one folder, the auto-set simIDs could not support "parfor". We may need another effective way to fix this promble.

…sim-id additional options for controlling simID and the main output file

make the main function return output and used params

…case/WiLabV2Xsim into dev-parallel-and-tracefile

…e/WiLabV2Xsim into dev-parallel-and-tracefile

Pos delay enhancement

V2Xgithub · 2024-04-22T09:42:22Z

Thanks Chester for your very appreciated contribution. We will do our best to look at your code, although we can't promise to process it as soon as it would deserve. A comment from my side is that, assuming that I correctly understood your text, the functionality you are describing was in principle already implemented in the simulator. Indeed, for this scope some of the vectors/matrixes use indexID and others directly the ID. When the indexID is used, this is because not all consecutive IDs may be active and we need to keep some matrixes as small as possible. As an example, the matrix including the channel gain has a number of rows and columns equal to the "active" users and not the "possible" users. The addition and removal of IDs is managed within the function "updatePositionFiles". Alessandro Bazzi, Ph.D. Associate Professor @ DEI, Università di Bologna Associated with WiLab, CNIT www.alessandrobazzi.com<http://www.alessandrobazzi.com/> Phone: +39-051-2093880

…

________________________________ Da: Chester Koh ***@***.***> Inviato: lunedì 22 aprile 2024 09:10 A: V2Xgithub/WiLabV2Xsim ***@***.***> Cc: Subscribed ***@***.***> Oggetto: [V2Xgithub/WiLabV2Xsim] [Draft] Basic Tracefile with Dynamic Vehicle Entry/Exit + some other fixes (PR #37) Hi V2XSim Maintainers, I have a working version of the sim with dynamic vehicle entry/exit partially supported for you to review. The approach/constraints are as follows: * The maximum number of active vehicles throughout the simulation is set constant at the beginning of the simulation. * The vehicle ids in the tracefile must be consecutive integers from 1 to XXX (use a python script to preprocess the bologna tracefiles) * A rectangle bound in world coordinates is specified using XminTrace, XmaxTrace, YminTrace, YmaxTrace * To mark a vehicle as active, the tracefile has a log line that shows the vehicle within these bounds * When a vehicle exits from the simulation, the tracefile adds a log line in the next epoch that "teleports" the vehicle outside these bounds * Vehicles "inactive" stay in some arbitary place outside of the active rectangle bound and do not request BR etc. They still tick for position updates and physics calculations (change in LOS/shadowing), but do not schedule any packets so the centralized algorithms do not schedule them any BR. * For the vehicle to become active again, the tracefile adds another log line to teleport the vehicle back within these bounds Using this approach I have a rudimentary support of dynamic vehicle entry/exit. The ideal scenario would be to exclude inactive vehicles from physics calculations entirely to speed up the simulation, but currently it is good enough for my current studies that requires tracefiles. My studies is specifically 5GNR centralised mode, so I will need your help to review the PR to test the other modes (decentralised, and maybe vehicular wifi) This PR contains some other files i accidentally committed from many months ago like MatFilesVisualization for my own studies but not fully functional by themselves yet. But the bulk of the changes that supports the tracefile is in the last commit. Attached is the modified BolognaAPositions with the constraints (consecutive vehicle numbers, and "teleporting" inactive vehicles to coordinates 0,0 ), preprocessed using Python. BolognaAPositions.csv<https://github.com/V2Xgithub/WiLabV2Xsim/files/15058862/BolognaAPositions.csv>

chestnutcase · 2024-04-22T12:15:43Z

😅 thanks for your prompt reply
the bologna tracefiles were not working out-of-the-box, and from previous discussion with felix i thought full support for tracefiles was abandoned so i was working on getting it working again. i guess the functionality change my version has is that the same vehicle id can re-enter the simulation after exiting. - from trying to reverse engineer the existing codebase, it seems like the original idea was that scenario exit can only happen once.

in any case, i will continue optimizing the code for full tracefile support because i need it for my studies. yes, I am aware that some of the data structures like channel gain uses matrix indices - the challenge now is to find these matrices that will require resizing when vehicles enter/exit. first i will work on removing exited vehicles from physics ticks (e.g. channel gain calculation) , then i will look into using tables instead of matrices (recent MATLAB 2023 feature: using tables in place of matrices) so that we can use arbitrary string indices everywhere (since row names are strings). It should be faster and more maintainable than trying to remember which data structure is indexed using what.

chestnutcase and others added 17 commits September 27, 2023 20:53

add output parameter to suppress printing to main file, and add optio…

8d1b816

…n to manually specify simID

make the main function return output and used params

09f55f0

Merge pull request V2Xgithub#24 from chesnutcase/print-main-file-and-…

f2d63c8

…sim-id additional options for controlling simID and the main output file

Merge pull request V2Xgithub#25 from chesnutcase/main-returns-outputs

fa34ab3

make the main function return output and used params

add support for counting hidden node events, and add unit tests

d7ad5ba

fix typo in test case for hiddenNode, should be FD not HD

ad5de75

attempt to support a configurable buffer size

5d86130

merge

2279a95

Merge branch 'configurable_buffer_size' of https://github.com/chesnut…

356227e

…case/WiLabV2Xsim into dev-parallel-and-tracefile

enhance usage of posDelay, add posPacketLoss param

6f48040

fix typo in class member of constants class

4e99b43

Merge branch 'pos_delay_enhancement' of https://github.com/chesnutcas…

2296293

…e/WiLabV2Xsim into dev-parallel-and-tracefile

fix wrong bounds check for simParams.posPacketLoss

6511566

fix wrong constant name

8fa1a46

Merge pull request #1 from SUTDFCCLab/pos_delay_enhancement

6524130

Pos delay enhancement

rudimentary support for dynamic vehicle entry and exit in traffic trace

8dcbdcb

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

[Draft] Basic Tracefile with Dynamic Vehicle Entry/Exit + some other fixes #37

[Draft] Basic Tracefile with Dynamic Vehicle Entry/Exit + some other fixes #37

chestnutcase commented Apr 22, 2024

V2Xgithub commented Apr 22, 2024 via email

chestnutcase commented Apr 22, 2024

[Draft] Basic Tracefile with Dynamic Vehicle Entry/Exit + some other fixes #37

Are you sure you want to change the base?

[Draft] Basic Tracefile with Dynamic Vehicle Entry/Exit + some other fixes #37

Conversation

chestnutcase commented Apr 22, 2024

V2Xgithub commented Apr 22, 2024 via email

chestnutcase commented Apr 22, 2024