Rocky 09-08-11-Ant Colony

download Rocky 09-08-11-Ant Colony

of 36

Transcript of Rocky 09-08-11-Ant Colony

  • 8/4/2019 Rocky 09-08-11-Ant Colony

    1/36

    ANT COLONYANT COLONYALGORITHM ANDALGORITHM AND

    ITSITS

    APPLICATIONSAPPLICATIONS

    Submitted by,ROCKEY SUSEELAN

    M.Tech(2nd SEM), E.S

  • 8/4/2019 Rocky 09-08-11-Ant Colony

    2/36

    CONTENTSCONTENTS

    INTRODUCTION.

    BEHAVIOR REAL OF ANTS.

    ACO.

    APPLICATIONS.

    ADVANTAGES.

    DISADVANTAGES.

    CONCLUSION.

    REFERENCES.

  • 8/4/2019 Rocky 09-08-11-Ant Colony

    3/36

    INTRODUCTIONINTRODUCTION

    The Ant Colony Algorithm is based on thebehavior of real ants foraging for food.

    Real ants communicate with each otherusing an aromatic essence calledpheromone, which they leave on the pathsthey traverse.

  • 8/4/2019 Rocky 09-08-11-Ant Colony

    4/36

    Natural behavior of ants have inspired

    scientists to mimic insect operational

    methods to solve real-life complex

    problems.

    By observing ant behavior, scientists have

    begun to understand their means of

    communication.

    Ant-based behavioral patterns to address

    the real time problems - first proposed by

    Marco Dorigo.

    REAL ANT BEHAVIOR

  • 8/4/2019 Rocky 09-08-11-Ant Colony

    5/36

    ANT BEHAVIOR

    The more ants follow a trail, the more attractive thattrail becomes for being followed

  • 8/4/2019 Rocky 09-08-11-Ant Colony

    6/36

    EXPERIMENTAL STUDY OF ANTS

    The more ants follow a trail, the more attractive thattrail becomes for being followed

    NEST FOOD

    NEST FOOD

    NEST FOOD

  • 8/4/2019 Rocky 09-08-11-Ant Colony

    7/36

    Let um and lm be the number of ants thathave used the upper and lower branches.

    The probability Pu(m) with which the(m+1)th ant chooses the upper branch is:

    )()()()(

    klku

    kuP

    mm

    mm

    hh

    h

    u

    !

  • 8/4/2019 Rocky 09-08-11-Ant Colony

    8/36

    ROUTE SELECTIONROUTE SELECTION

    Ants are forced to decide whether they should

    go left or right, and the choice that is made is a

    random decision.

    Pheromone accumulation is faster on the

    shorter path.

    The difference in pheromone content between

    the two paths over time makes the ants choose

    the shorter path.

  • 8/4/2019 Rocky 09-08-11-Ant Colony

    9/36

    GOAL OF ACO HEURISTICGOAL OF ACO HEURISTIC

    Artificial ants form a multi-agent system

    performing the functions as observed in the

    real ant system.

    The ACO meta-heuristic relies on the co-operation of a

    group of artificial ants to obtain a good solution to a

    discrete optimization problem such as the TSP

    Artificial ants are mutants of a real antsystem.

    The resulting shortest route mappingdetermined by the agents can be applied tothe optimization problem.

  • 8/4/2019 Rocky 09-08-11-Ant Colony

    10/36

    REAL vs. ARTIFICIAL ANTSREAL vs. ARTIFICIAL ANTS

    Discrete time steps

    Memory Allocation

    Quality of Solution

    Time of Pheromonedeposition

    Distance

    Estimation

    REAL ANT ARTIFICIAL ANT

  • 8/4/2019 Rocky 09-08-11-Ant Colony

    11/36

    ACOCHARACTERISTICS

    Exploit a positive feedback mechanism.

    Demonstrate a distributed computational

    architecture.

    Exploit a global data structure that changes

    dynamically as each ant transverses the route.

    Involves probabilistic transitions among states

    or rather between nodes.

  • 8/4/2019 Rocky 09-08-11-Ant Colony

    12/36

    FLOWCHARTOF ACO

    Have allplace been

    visited

    Have themaximum

    Iterations beenperformed

    START ACO

    Locate ants randomlyin places across thegrid and store the

    current place

    in the list

    Determine probabilisticallyas to which place to visit next

    Move to next place andput this place in the

    list

    Record the length oftour and clear the list

    Determine the shortesttour till now and

    update pheromone

    NO

    YES

    STOPACO

    YESNO

  • 8/4/2019 Rocky 09-08-11-Ant Colony

    13/36

    KEY PARAMETERS

    Trail intensity is given by value ofXij which indicates the

    intensity of the pheromone on the trail segment.

    Trail visibility (Heuristic factor) is Lij = 1/dij .

    The relative importance of the intensity in theprobabilistic transition is E.

    The importance of the visibility of the trail segment is F.

    The trail persistence or evaporation rate is given as V.

    Q is a constant and the amount of pheromone laid on atrail segment employed by an Ant. This amount may be

    modified in various manners.

  • 8/4/2019 Rocky 09-08-11-Ant Colony

    14/36

    PROBABILISTIC SELECTION

    Helps determine the city to visit next while

    the ant is in a tour.

    Determined by variables such as the

    pheromone content in an edge (i,j) at timeinstant t, heuristic function of the desirability

    of adding edge, and their control parameters.

    ? A ? A? A ? A

    )(

    )(

    0

    )()(

    )()(

    iJjif

    iJjiftt

    tp

    k

    kilil

    ijij

    k

    iji

    kJl

    !

    FE

    FE

    LXLX

  • 8/4/2019 Rocky 09-08-11-Ant Colony

    15/36

    ALGORITHMS3 ALGORITHMS

    The ant-cycle is the approach discussed so far

    Information is updated at the end of each tour as such

    function of tour length

    The ant-density is an approach wherein the

    pheromone quantity Q is deposited once the segment

    is transverse.

    Pretty much a greedy approach (local information) and not

    really providing relative information

    The ant-quantity is an approach wherein thepheromone quantity Q/dij is deposited once the

    segment is transverse.

    Also a greedy approach but providing some relative

    information by scaling Q by the length of the segment

  • 8/4/2019 Rocky 09-08-11-Ant Colony

    16/36

    TRAVELLING SALESMANPROBLEM

    OBJECTIVE:

    Given a set of n cities, the Traveling

    Salesman Problem requires a salesman tofind the shortest route between the given

    cities and return to the starting city, while

    keeping in mind that each city can be visited

    only once.

  • 8/4/2019 Rocky 09-08-11-Ant Colony

    17/36

    TRAVELLING SALESMAN PROBLEM

    Ant Cycle Algorithm

    Step 1:

    pheromone traces have equal values for every edge.

    place ants in graphs nodes.

    Step 2: locate start-up node (city) of every ant on first

    position in list.

    Step 3:

    repeat, for every ant, until list is full move ant from current node to a neighbour node,

    chosen with probability

    depending on traces value and on weight of edgebetween the two nodes -/

    put chosen node in ants list

  • 8/4/2019 Rocky 09-08-11-Ant Colony

    18/36

    Step 4: For each ant,

    compute tour length described by the ant. modify the optimal value found (in case of a better

    value).

    add pheromone on tours edges, depending of thislength.

    Step 5:

    increment number of iterations.

    Step 6:

    if (maximum iterations number is not reached) and

    there is no stagnation behaviour)

    then

    empty listsgo to step 2

    else

    print best solution found

    end algorithm

  • 8/4/2019 Rocky 09-08-11-Ant Colony

    19/36

    TSP APPLICATIONS

    Lots of practical applications

    Manufacturing routing such as movement of

    parts along manufacturing floor or the amountof solder on circuit board.

    Network design such as determining the

    amount of cabling required

    Two main types

    Symmetric

    Asymmetric

  • 8/4/2019 Rocky 09-08-11-Ant Colony

    20/36

    SYMMETRICAL FPGA PLACEMENT

    The Field Programmable Gate Array (FPGA)

    has gained rapid commercial acceptance as

    of its low cost and fast prototyping.

    But the area efficiency and performance are

    still their drawbacks. So it is necessary to

    apply efficient algorithms to handle FPGA

    placement for obtaining good area efficiency

    and performance.

  • 8/4/2019 Rocky 09-08-11-Ant Colony

    21/36

    FPGA

    All FPGAs are composed of three fundamental

    components: logic blocks, I/O blocks and

    programmable routing resource.

    A circuit is implemented by programming each of

    logic blocks to implement a small portion of the logic

    required by the circuit.

    Each of the I/O blocks acts as either an input pad or

    an output pad. The programmable routing resource is

    configured to make all necessary connections among

    different logic blocks. Since routing resources may

    occupy 70-90% of the chip area, a good placement

    and routing algorithm is very important to overall

    FPGA design.

  • 8/4/2019 Rocky 09-08-11-Ant Colony

    22/36

    FPGA ARCHITECTURE

    Each routing channel is assumed to have a fixed

    number of channel tracks.

    Each wire segment spans the distance of one

    CLB. Two IOBs fit in the space of each CLB along

    the periphery of the FPGA.

  • 8/4/2019 Rocky 09-08-11-Ant Colony

    23/36

    ACO PLACEMENT

    FPGA placement is a process which assigns the

    blocks in the net list of circuit into CLBs in FPGA

    chip.

    In the process, the position of the blocks can bechooses to move so that the cost function is

    minimized gradually. At the end of placement,

    we can get an optimal result.

    The most important component of using ACO

    algorithm in placement is the management of

    pheromone trails.

  • 8/4/2019 Rocky 09-08-11-Ant Colony

    24/36

    According to ACO, we can select the CLBs as the

    pheromone trail loader. We set an initial value of

    pheromone for each CLB. The value of pheromone

    trail will update with the process of placement going

    on

    The pheromone trail according to the following

    formula:

  • 8/4/2019 Rocky 09-08-11-Ant Colony

    25/36

    In the above formula, ij represents the phremone

    value of assign Blockj into CLBi. And represents

    evaporation coefficient in ACO algorithm.

    In the process of placement, if CLBi is selected by the

    Blockj in the current placement, we will update the

    pheromone trail of CLBi.

    We will compute the value of ij according to the

    following formula:

  • 8/4/2019 Rocky 09-08-11-Ant Colony

    26/36

    In the above relation, cur_cost is the cost value of

    current placement, Q is user-defined parameter.

    Since other CLBs except CLBi

    are not selected in

    this placement, we only decrease the pheromone trail

    for them. That is to say, ij is equal to zero for

    these CLBs in this placement.

    The probability for these CLBs in the next iteration is

    large and other CLBs are almost ignored. It will make

    the proposed algorithm a local search algorithm.

  • 8/4/2019 Rocky 09-08-11-Ant Colony

    27/36

    In order to avoid the situation, we introduce a new

    technique named Max-Min Ant System(MMAS).In this system,

    a max value and a min value will be given for pheromone trail.

    In first formula

    min and

    max represent the min value andmax value of pheromone trial in the stage of placement.

    In the routing stage, the channel width is decided by the

    max width of channels. So if we consider the overlapping

    among the bounding boxes in the stage of placement, thechannel width of all channels will be decreased. Then the

    routing congestion will also be reduced in the stage of routing.

  • 8/4/2019 Rocky 09-08-11-Ant Colony

    28/36

    The probability of assign Blocki to CLBj

    depends on the combination of pheromone

    trail and attractiveness. We can compute it

    according to the following formula:

    where tabuk is the tabu list of ant k, while parameters andspecify the impact of trail and attractiveness, respectively.

  • 8/4/2019 Rocky 09-08-11-Ant Colony

    29/36

    ADVANTAGES

    Parallel Computation Mechanism.

    The Positive Feedback in ACO accounts for rapid

    discovery of good solutions.

    ACO employs Distributed computation, which

    avoids premature convergence.

    Easy Combination With Other Methods.

    Good Robustness.

  • 8/4/2019 Rocky 09-08-11-Ant Colony

    30/36

    DISADVANTAGES

    These algorithms have slower convergence thanother Heuristics.

    They performed poorly for TSP problems larger

    than 75 cities.

    Long-time Searching.

    There is no centralized processor to guide the

    Ant System towards good solutions.

  • 8/4/2019 Rocky 09-08-11-Ant Colony

    31/36

    CONCLUSION

    As a structure-based algorithm essentially for solving thediscrete domain optimization problems, ant colony

    algorithm has many merits, but at the same time it also

    has shortcomings including a long searching time and the

    local optimum.

    Because of its promising characters such as strong

    robustness powerful searching ability and easy

    combination with other intelligent algorithms, it provides

    wide range of application.

  • 8/4/2019 Rocky 09-08-11-Ant Colony

    32/36

    ReferencesReferences

  • 8/4/2019 Rocky 09-08-11-Ant Colony

    33/36

    ReferencesReferences

  • 8/4/2019 Rocky 09-08-11-Ant Colony

    34/36

    ReferencesReferences

  • 8/4/2019 Rocky 09-08-11-Ant Colony

    35/36

    ReferencesReferences

  • 8/4/2019 Rocky 09-08-11-Ant Colony

    36/36

    ReferencesReferences