readme.txt

H. Yang, A.P. Petropulu, X. Yang and T. Camp, ?ALLIANCES with Optimal Relay Selection,"IEEE Trans. o

Improtant notes:
(1) the comments are sufficient in code. please read the comment carefully before any modification of code.
(2) since most programs are simulation result (items 2-6), the codes involve extensive Monte Carlo simulations. The BER is typically smaller than 10^-3 thus running times of the Monte Carlo simulations M must be sufficient large. In order to get non-zero BER at hign SNR case, it is recommended that M = 10^6 or even larger. This process will take around 10 hours or even longer, depending on the computer performance.
(3) a channel matrix must be generated first. the matlab data file channel_ng_k5b.mat was generated using channel_gen_ng.m (in alliances.zip) with proper parameter settings.
Matlab m-files for the optimal relay paper
1. generating theoretical result of BER vs SNR for K=4,l=0,1,2,3 and K=1, or other (K,l) pairs. Using petest_main_dist.m as main function. petest_dist.m and pecompute_dist.m and src_gen.m and ei_1.m are invoked by the main m-file.
The output figure here is named as BER_SNR_K4_new_theo.fig/.eps

2. generating simulation result of BER vs SNR for K=4,l=0,1,2,3, or other (K,l) pairs. Using coop_k_ber0606_K4_simu.m as main function. 
The output figure here is named as BER_SNR_K4_new.fig/.eps

3. generating simulation result of BER vs SNR for non-cooperative protocols K=1, It is very similar to program above. Using coop_k_ber0606_K1_simu.m as main function. 
The output is plotted in the same figure above as BER_SNR_K4_new.fig/.eps

4. generating BER vs Distance Rang r_out. Note that signal power and noise power and inner radius of ring area r_in are all fixed. Using coop_k_ber0606_vsdist.m as main function. 
The output figure is named as BER_vs_DistRange.fig/.eps

5. generating BER vs SNR for comparison of two relay selection schemes (shown improvement) for K=4. Using coop_k_ber0606_K4_RS.m as main function. Please note that in the very beginning of this main program if you set OptFlag = 0, the output will be BER for the simplest relay selection; if OptFlag=1 then it outpus BER for the optimal relay selection.
The output figure here is named as BER_K4_dist_RS.fig/.eps


6. generating BER vs Traffic Load for comparison of two relay selection schemes (show improvement). Using coop_p_ber0606_RS.m as main program. Please note that in the very beginning of this main program if you set OptFlag = 0, the output will be BER for the simplest relay selection; if OptFlag=1 then it outpus BER for the optimal relay selection.
The output figure here is named as BER_highSNR.fig/.eps

7. Introductions to other m-files.
decodpac.m and spheredecod3: the Maximum Likelihood decoder
mapper2.m: mapping the decoded result and compare it with originally transmitted packets.
channel_ng_k5b.mat: the channel coefficients matrix generated by the Jake's model. It is a slowly varying channel involving J=16 users. All user-BS and inter-user channels have unit variance and zero mean. Each channel contains 10^3 time slots.

petest_dist.m pecompute_dist.m src_gen.m ei_1.m are used by petest_main_dist.m to compute theoretical BER for different SNRs.