[Naser]

Dear Fjolla/Matrin

When I run a demo (RRH demo — RRH disabled (false) for example) in “tri sector” scenario:
in The UE trace, I see UEs with certain TP and SINR .. these stats is the same If I re-run the same model again.
i.e. UEs have the same SINR and TP when i run it again (UEs are in the same position).

However, this is NOT the case in six sector scenario.. when I re-run the model.. UEs will have different stats (as UEs are in different positions).

How can I guarantee the same stat (same UE position when make a second run) for six sector , just like tri sector ??

Thanks,
Naser

[Fjolla Ademaj]

Dear Naser,

If I understood you correctly, you want to fix the user positions over realizations. In such case, if your users are randomly distributed, you can fix the seed such that in each call it gives you the same positions.

Best,
Fjolla

[Naser]

Dear Fjolla,

Thanks for your feedback, Yes, this is my question.

As per your note, I think the issue is in “constantElementsPerCellSpatialDistribution”
But I couldn’t do it, could you please help.

Secondly, why in tri sector, the UsE positions are the same for every realization ?

Thanks
Naser

[Naser]

Dear Fjolla,

Also, When I tried a fixed distribution (radial).
In tri-sector, two run (realizations) yield same results.. whuch is Ok.

But, when I changed to six sector, it is strange,
It yields same position, same SINR, but different throughput , as the (average_energy_per_bit) is different between two realizations (even the SINR is the same !

Any suggestions please.

I need two realization yiels same result (as the case in tri_sector).

Thanks,
Naser

[Fjolla Ademaj]

Dear Naser,

Regarding your first post: in “constantElementsPerCellSpatialDsitribution”, the user positions are randomly generated by using the randi function, thus in each realization you will get different user positions.

There is an option LTE_config.UE_cache, if it is set to true, then in the first simulation the UE positions are saved in a file and in the rest of realizations the user postions are read from that file. This might be the case for you when you simulate with the tri-sector setup.

Regarding your second post: Try to check your parameter settings, if they are the same as in your tri-sector case. Check the LTE_config.UE_cache parameter, and also check if you save or not the channel trace, i.e. LTE_config.delete_ff_trace_at_end (set to ‘true’ means that the channel trace is deleted at the end of the simulation) and LTE_config.recalculate_fast_fading (with this option set to ‘true’, even if cache is present, the channel trace will be recalculated).

To get the same result, you need to use the same channel trace.

Best,
Fjolla

[Naser]

Dear Fjolla,

Thanks for the input,

I have set the following:

LTE_config.recalculate_fast_fading =0;
LTE_config.UE_cache= true;
LTE_config.delete_ff_trace_at_end= false;

This works only for tri sector
Yet, the sex sector scenario is still the same, UEs in different positions compared with the previous realization, the peak/avg/edge TP is also different.
Any suggestions why the six_sector_tilted is having this behavior ?

Best regards,
Naser

[Naser]

Dear Martin,

Any suggestions why the six_sector_tilted is having this behavior ?

How can we keep UEs position in two realizations ?

best regards,
Naser

[Martin Müller]

Hi Naser,

you need to make sure that the cache file that you created with the first simulation (i.e., the tri-sector simulation) is also actually used for placing the UEs in the second simulation (i.e., the six-sector simulation). Please refer to the function LTE_init_generate_users_and_add_schedulers.m. There, it is checked if a cache file exists with a specific name. By default, the file-name of the stored UE-cache is generated automatically (see LTE_load_params_dependant.m and +utils/naming.m). If you specify your own file-name by setting LTE_config.UE_cache_file = ‘my_cache_file’ for both simulations, it will be created in the first simulation (since it doesn’t exist) and will be stored and reused in the second one.

Best regards,
Martin

[Author]

Posts