CMPUT 313 Assignment 2
Stephen Jahns and Marek Kundera

DISCUSSION
    Steve: In main.cpp, I don't think we need to print the seeds:
    "The output consists of several lines. The first line of output repeats the
    input parameters (without the seeds), again formatted in one single line."

TODO - 
    Steve - TDMStation, PBStation
    Marek - IBStation TBEBStation

Goal - Compare different backoff strategies for Slotted ALOHA MAC layer protocol
     - ALSO, compare ALOHA strategies with TIME DIVISION MULTIPLEXING
     - Comparisons based on THROUGHPUT and AVERAGE DELAY
        - AVERAGE DELAY is full delay between frame generation and completed transmission?
                - measured in slot times required for successful tx of a frame
        - THROUGHPUT is (# frames tx'd X frame size in bits) / time in slots ???


Problem - BUS with N stations
        - Assume frames of fixed, equal size
        - Transmision time for one frame exactly equal to a single slot time
        - Beginning of slot synced to all stations simulaneously (no prop. delay)
        - COLLISION when 2 or more stations transmit frame in same slot (on same bus)
            - collision renders the transmissinos corrupted/unrecoverable
            - collision INSTANTLY DETECTED (at beginning of slot...) by all stations
        - Each station generates new frames independantly
            - Generation controlled by global parameter 'p' - probability that station
             will generate a new frame in a slot - shared by all stations
            - Generated frames added to a queue of untxed frames
            - Head of queue considered for transmission at each time slot

Protocols to evaulate:
    PROTOCOL T - Time Division Multiplexing
               - Stations numbered 1 to N.
               - Tranmission in cycles of N slots
               - Slots numbered starting at 1
               - Station 'i' allowed to transmit in slot mN+i, m = 0, 1, 2, ... 
                 where m is cycle number
               - Time slot wasted if station has nothing to tx

    PROTOCOL P - Slotted ALOHA with probablistic backoff
               - Station with frame to tx will tx ASAP (may be in same slot as generation)
               - If collision on first attempt
                    -TX in next slot with prob 1/N
                    - If collison, tx in next slot with prob 1/N ... and so on till success

    PROTOCOL I - Slotted ALOHA with interval based backoff
               - Station with frame to tx will tx ASAP, (may be in same slot as gen)
               - If collision on first attempt:
                    -Next attempt performed in slot uniformly chosen between 1 and N after 
                    current slot.
                    -Attempt until success

    PROTOCOL B - Slotted ALOHA with truncated binary exponential backoff
               - Station with frame to tx will tx ASAP, (may be in same slot as gen)
               - If collision on first attempt:
                    - next attempt performed in 1st, 2nd, 3rd, or 4th slot, randomly
                    - If still collide, backoff interval expands geometrically (doubling)
                        - 1st - 4 slots, 2nd - 8, 3rd - 16
                        - UNITL backoff interval is 1024 possible slots - CAP, interval
                        used until tx is successful
                - No memory of backup interval - next frame transmission will always occur 
                in next frame after successful transmission


Simulation
    - Program parameters:
        psim Protocol N p R T seed1 seed2 ... seedT
        where 
            - 'Protocol' is the name selected from {'T', 'P', 'I', 'B'}
            - 'N' is the number of stations sharing the bus
            - 'p' is the grame generation probability for each station
                - note: consider p <= 0.04
            - 'R' is the total number of slots that need to be simulated
                - note: R=50000 should be good
            - 'T' is number of trials, followed by the RNG seeds for each of the T trials

    - Output:
            - Line 1: repeat input parameters (without seeds)
            - Line 2: Average throughput of system followed by lower, upper bounds of 95% CI
                      calculated over all T trials.
            - Line 3: Overall average per-frame delay in slot times follwed by 95% CI 
                        note: lowest possible delay is 1 slot
            - N * 4 subsequent lines: performance with respect to each station
                      Line 1: node ID
                      Line 2: station throughput + CI
                      Line 3: station delay + CI
                      Line 4: Ratio of undelivered frames to total generated frames at end of
                              sim FOR EACH OF T trials (dont average)


DESIGN:

    Main method - creates instance of Simulator as per command line argument,
                  runs it

    Simulator - contains code common to each of the 4 protocols 
              - Contains a vector of 'Stations' - representing all the
              stations on the bus

    -> Subclass Station for each protocol?
        TDMStation - T
        PBStation - P
        IBStation - I
        TBEBStation - B
       
        - Each station type will handle transmission eligibility and
        collission handling differently

    Running experiments - Use similar bash scripts connected to GNUPlot as for
    PSIM project

    What does the simulator need to DO? -- how can we exploit polymorphism to
    make things easy...

    For each of T trials...
        Simulate events occuring over R timeslots...
        During each timeslot,
            For each station, 
                generate a frame with probability p
                Determine whether or not a frame tx should be attempted 
                    -> Here is where polymorphism occurs 
                        - TDM will transmit if the slot number corresponds to the
                        station
                        - ALOHA will tx immediately IF first tx OR last tx attempt 
                        successful, else will tx according to whatever backoff algorithm
                        its particular instance uses
            Check how many stations tried to transmit -> If only one, remove
            the txed frame from its tx queue (no collision!)

        At trial end, determine per station ratio of undelivered to total
        generated frames -> store in some vector or structure
        Determine overall throughput for trial, store in a vector
        Maintain running average of overall per-frame delay? also for each
            station

    After all trials
    -> Determine average throughput with CI
    -> Determine average per-frame delay with CI
    -> Determin per station throughput, delay