User talk:Ywa ssd imr

This page will be used for discussion on the usage of the simulator. If you have any questions regarding the simulator, please add it below. I'll then try to anser it.

E.g. Q: How should I use this simulator? A: To start with, go into the folder sys_link_simu, and excute the following commond from matlab: sys_link_simu_func(a,b) where a is the link adaptation interval, in terms of No. of OFDM frames. b is the power adaptation interval. e.g. a = b = 1. '''Q: I found it very interesting and it would be helpful if you could provide me more information about the system level simulator architecture, in the direction of mapping table from link level. ''' A: The most essential assumption that we took, is that the multi-cell interference has similar impact on the signal detection as white Gaussian noise. Since each user received 18 interferences from the 18 neighboring cells (at least 6 will be very strong interference sources, from cells in the first tier), this assumption can roughly be met. Moreover, OFDMA is used as the access technique, so there's no interference from other users. Indeed, we have the PDF, CDF, and level crossing rate plot for SNR (with only Gaussian noise), and the measured SINR (Gaussian + Multi-cell interference), they are matching pretty fine with each other.

With this understanding, a given SINR (in multi-cell system level simulator) is expected to offer similar performance with the same SNR value (in single cell link-level simulator).

We tries to satisfy a certain Block Error Rate (BLER) target, while aiming to improve the spectral efficiency, fairness etc. Thus in the system-level simulation, after measuring the received SINR and feed it back to NB, we can map it to BLERs using the mapping table generated in the link-level simulator (different BLER for different coding & modulation level). The proper coding rate and modulation level can be choosen by comparing the achievable BLERs with the BLER target. The feedback will take some time, and this could lead to very different performance, if not handled properly, in a high mobility scenario. We use an additional SNR margin to guarantee the BLER, means more power is assigned. This margin is found in the link-level simulator automatically.

By using this approach, we avoid the high calculation complexity in the system level simulation, as it is next to impossible to do the 'real' simulation for each UE in a multi-cell multi-user multi-antenna multi-rate multi-others system.

In short, the mapping table is the interface between the link-level simulator and the system-level simulator. It maps the SINR to a certain achievable BLER.

'''Q: How long does it take to run a simulation? I have been running for hours and it still didn't end.Also, how does one plot other plots like spectral efficiency for link level simulation? How are directories organized? How does one load the corresponding MAT files and what to do with them?'''