A realistic bulk service system

Implementation of a bulk system simulator for the course Statistical modelling and design of experiments of 2015/16.

Instructions

To build:

$ make             # Uses the exponential distribution for service times
$ make -dUSE_NORM  # Uses the normal distribution for service times

To run:

$ bin/simulator -f <input file>
               [-s <seed>] [-o <output>]
               [-c <cycles>] [-r <rho>]
               [-v] [-h]

The input file has to be in the following format:

E[tau]
PD(tau)
E[x]
PD(x)
E[c]
PD(c)
C
mu
Amax

Where PD(x) is the probability distributions used for variable x. The supported distributions are the following (values in <> are numbers):

CONS <num>            # Constant value
UNIF <a> <b>          # Uniform distribution
NORM <mean> <sd>      # Normal distribution
EXP <rate>            # Exponential distribution
ERLANG <shape> <rate> # k-Erlang distribution

To get a representation based on the trace file run the following command (Note: R must be installed):

$ make images/sth.png    # Assumes traces/sth.csv exists

You can also run these scripts directly:

$ utils/script.R in.csv out.png    # Used for the command above 
$ SEEDS=`seq 1 10` CYCLES=100 utils/table.sh  # Generates a table with values for different seeds

The structure of the project is the following:

src/       Source files
build/     Build files  
bin/       Location of executable
utils/     Utility scripts
data/      Sample input data
traces/    Location of execution traces
images/    Visual representation of traces