READ ME File For 'two_dimensional_IM_dataset'

Dataset DOI: 10.5258/SOTON/D1000

ReadMe Author: SIYAO LU, University of Southampton 

This dataset supports the publication:
Siyao Lu, Mohammed El-Hajjar, Lajos Hanzo
Two-Dimensional Index Modulation for the Large-Scale Multi-User MIMO Uplink
IEEE Transactions on Vehicular Technology


This dataset contains:
The data used for producing Figures 3, 4, 5, 6, 7, 8, 9, 10, 11, 12. 
Each folder is named according to the figure's number and the dataset - of each figure - is stored in one or two .dat file in that folder. 
There are THREE different plot types: BER plots, Maximum Achievable Rate plots and SNR-penalty plots.
All figures can be re-created by using GLE with the aid of .dat files of each figure, where each .dat file represents each curve in a specific figure. 
Explicitly, you should first write a .gle file using the .dat files of a specific figure, where you can decide the size, the color, or the caption of the figure depending on your requirement. 
Then you should run the .gle file by typing "gle filename.gle" in your terminal and the figure is re-created.

The figures are as follows:

Fig. 3  Comparison between the uncoded BER performances of Scheme 1, Scheme 2, Scheme 9 of [11] and Scheme 10 of [18] shown in Table V using ML detectors, at the same rate of Rt = 1.88 bits/s/Hz/user. 

Fig. 4  Comparisons between uncoded BER performances of Scheme 2 shown in Table V using both the MMSE detector and the proposed RSS-IMP MUD with I = 1 to 3 iterations applied, at the same rate of Rt = 1.88 bits/s/Hz/user.

Fig. 5  Comparison between the uncoded BER performances of Scheme 11 using both the ML detector and the proposed RSS-IMP detector, of Scheme 12 using the ML detector and of Scheme 13 using the ML detector, while supporting the same number of U = 3 users, the same number of RAs Nr = 36 and the same rate of Rt = 2.5 bits/s/Hz/user. 

Fig. 6  Comparison between the uncoded BER performances of Scheme 3 and Scheme 4 for LS-MU-MIMO-UL scenario using both the MMSE detector and the proposed RSS-IMP detector having I = 1 to 3 iterations.

Fig. 7  Comparison of the SNRs required at the BER target of 10−4 with varying values of the RAs at the BS for Scheme 8 using both the MMSE detector and the RSS-IMP detector using I = 3 iterations and for Scheme 20 using the ML detector.

Fig. 8  SNR-penalty wrt the perfect CSI scenario at the BER target of 10−4, where Scheme 5, Scheme 6 and Scheme 7 shown in Table V are employed and we have varying values of channel estimation error variance σh^2 .

Fig. 9  Maximum achievable rates of Scheme 14, Scheme 15 and Scheme 16 in MU scenarios applying the ML detector.

Fig. 10 Maximum achievable rates of Scheme 17, Scheme 18 and Scheme 19 in LS-MU scenarios applying the RSS-IMP detector using I = 4 iterations.

Fig. 11 The average BER performances of the iteratively detected half-rate RSC-coded system shown in Fig. 1 based on the detection of (24) in conjunction with the parameters of Scheme 14 shown in Table V and an interleaver depth of 300,000 bits while using I_IO = 1 to 4 iterations.

Fig. 12 The average BER performances of the iteratively detected half-rate RSC-coded system shown in Fig. 1 based on the detection of (36) using I = 4 iterations in conjunction with the parameters of Scheme 17 shown in Table V and an interleaver depth of 300,000 bits while using I_IO = 1 to 4 iterations.


Date of data collection: 02-07-2019

Information about geographic location of data collection: University of Southampton, U.K.

Licence: CC-BY

Date that the file was created: July, 2019