readme.txt

无线仿真Matlab代码

Description of the Matlab pre-processing/generation of input trace files 
for the EURANE simulator of ns-2.

We have the following files:
----[main file]:--------------------------------------------------------
chan_generator.m 
----[function files in order of calling]:-------------------------------
environment.m    function [power,spectrum] = environment(multi,max)
multipath.m      function [omega] = omega(samples,rayleigh_fades,fD,spectrum)
shadowing.m      function [S] = shadowing(numTTI,shadow_std,alpha,beta)
Power1stRT.m     function [P_1stRT] = Power1stRT(numTTI,P_out,HARQcycle)
Power2ndRT.m     function [P_2ndRT] = Power2ndRT(numTTI,P_out,HARQcycle)
Power3rdRT.m     function [P_3rdRT] = Power3rdRT(numTTI,P_out,HARQcycle)
compCQI.m        function [CQI,perctoolow,perctoohigh] = 
                        compCQI(numTTI,P_out,CQIdelayinTTI,minCQI,maxCQI)
----[subroutine files]:--------------------------------------------------
parameters.m       called at the begin of chan_generator.m
savetrace.m        called at the end of chan_generator.m

Make sure to create an 'output/' directory where the output trace files 
are collected.

Parameters are divided in two groups:
1)parameters that may be changed by people interested in the network layer 
  and need this matlab pre-processing to generate input traces for ns-2.
  These parameters are requested during the actual matlab simulation by 
  means of a input dialog.
2)parameters that may be changed by people interested in the physical layer.
  These parameters are located in parameters.m

----The code is generating the following file (/output/ directory):
<env>-<vel_kmh>kmh-<distanceUEBS>m-<dur>s-UEnr<n>
with <env>          the environment defined in parameters.m
     <vel_kmh>      the velocity in km/h
     <distanceUEBS> the distance between UE and BS in metres
     <dur>          the duration of the simulation trace in seconds
     <n>            the index of the trace being generated
The result for example looks like:
Ray_corr-3kmh-500m-200s-UEnr1
Ped_A-3kmh-700m-100s-UEnr1
Ped_A-3kmh-700m-100s-UEnr2

It can be loaded in matlab as a matrix with the command:
load <filename>

Each trace file also contains the following pre-processing info:
(all lines starting with '%' such that matlab will discard it)
-physical layer parameters information 
-length of the trace file 
-the unique user identitity.
-directory in which it is created
-date at which it is created
-username of the person that generated it
-its original filename (handy in case it is renamed)
-HARQ and CQI information

The generation of traces consists of a loop over the user number, that 
starts with the definition of the random seed equal to the user number. 
In this way, traces can always be reproduced.
And the tracefiles of two users with equal specifications, will have the 
same statistics but not the same behaviour in time.

Note that the length of the trace should be at least as long as the time 
between start and finish for the event scheduler in ns-2. The latter is 
longer than the time interval of the traffic source.

Note that the first CQIdelayinTTI (3) elements and the last 2*HARQcycle (12)
elements are not written to the tracefile because they need power 
information which is outside the range being considered in the simulation.
During the matlab simulation, the tracefile is temporally made somewhat 
longer than the requested simulation length since somewhat more values 
should be generated within matlab than actually needed in the trace 
file output. This is due to the fact that the CQI at TTI 1 is based on 
earlier power levels and the retransmission powers need future power levels.
It may become more clear with the following schematic: 
  -2                       |  P       
  -1                       |  P    
   0                       |  P   
------------------------------------------------------------------------------------
   1                       |  P  P_1stRT  P_2ndRT  CQI       ||  Only this 
   ...                     |  P  P_1stRT  P_2ndRT  CQI       ||  part is 
   ...                     |  P  P_1stRT  P_2ndRT  CQI       ||  send to the
   numTTItrace             |  P  P_1stRT  P_2ndRT  CQI       ||  tracefile
------------------------------------------------------------------------------------
   numTTItrace+1           |  P    
   ...                     |  P    
   ...                     |  P    
   numTTItrace+2*HARQcycle |  P    


Mind the following:
-unit of time is the TTI which equals 2 milliseconds in HSDPA
-unit of speed is m/s in the internal part of the pre-processing while 
 it is km/h in the input dialog.
-Regarding the shadow fading, the result is translated and normalised 
 such that the resulting mean and standard deviation correspond to 
 0 and shadow_std.
 So, even for short simulations, two UE's that only differ in the
 random seed, do have the same statistics.