net.tomp2p.replication

Class AutomaticReplication

java.lang.Object

net.tomp2p.replication.AutomaticReplication

All Implemented Interfaces:

PeerInit, ReplicationFactor, PeerMapChangeListener

public class AutomaticReplication
extends Object
implements PeerMapChangeListener, ReplicationFactor

AutomaticReplication class observes the local network conditions and predicts the next departure count based on previous and current peer churn behaviors. This predicted value and given reliability are used to calculate the replication factor

Author:

Maxat Pernebayev

Constructor Summary

Constructors

Constructor and Description
AutomaticReplication(double reliability) Constructor.

Method Summary

Methods

Modifier and Type	Method and Description
int	calculateReplicationFactor(int predictedValue) The replication factor is calculated based on reliability and predicted value.
void	clearRemovedPeers() In order to track the number of removed neighbor peers every time, the hash set should be emptied for the next use.
int	factor()
Double	getAverage(ArrayList<Integer> observations, ArrayList<Double> averages) Exponential Moving Average with Dynamic Smoothing Factor
double	getBestSmoothingFactor(ArrayList<Integer> x, ArrayList<Double> y, int initialInterval)
int	getNeighbourPeersSize()
Double	getPredictedValue(ArrayList<Integer> observations, double currentAverage) Exponential Moving Average with Dynamic Smoothing Factor plus Deviation
int	getRemovedPeersSize()
double	getStandardDeviation(ArrayList<Integer> range, double currentAverage) Calculation of Standard Deviation for a set of values
double	getSumOfRegressionVariationSquared(ArrayList<Integer> x, int k, double b0, double b1, double yMean)
double	getSumOfXVariationMultipliedYVariation(ArrayList<Integer> x, ArrayList<Double> y, int k, double xMean, double yMean)
double	getSumOfXVariationSquared(ArrayList<Integer> x, int k, double xMean)
double	getSumOfYVariationSquared(ArrayList<Double> y, int k, double yMean)
double	getXMean(ArrayList<Integer> x, int k)
double	getYMean(ArrayList<Double> y, int k)
void	init(Peer peer)
void	peerInserted(PeerAddress peerAddress, boolean verified)
void	peerRemoved(PeerAddress peerAddress, PeerStatatistic storedPeerAddress)
void	peerUpdated(PeerAddress peerAddress, PeerStatatistic storedPeerAddress)

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Constructor Detail

AutomaticReplication

public AutomaticReplication(double reliability)

Constructor.

Parameters:

reliability - The reliability

peerMap - The map of my neighbors

Method Detail

init

public void init(Peer peer)

Specified by:

init in interface PeerInit

getNeighbourPeersSize

public int getNeighbourPeersSize()

Returns:

The number of neighbor peers.

getRemovedPeersSize

public int getRemovedPeersSize()

Returns:

The number of removed neighbor peers.

clearRemovedPeers

public void clearRemovedPeers()

In order to track the number of removed neighbor peers every time, the hash set should be emptied for the next use.

factor

public int factor()

Specified by:

factor in interface ReplicationFactor

Returns:

The replication factor.

calculateReplicationFactor

public int calculateReplicationFactor(int predictedValue)

The replication factor is calculated based on reliability and predicted value.

Parameters:

predictedValue - Predicted value from prediction model

Returns:

The replication factor

getXMean

public double getXMean(ArrayList<Integer> x,
              int k)

Parameters:

x - The array of independent values, i.e. observations

k - The start index of subset

Returns:

The mean of independent values

getYMean

public double getYMean(ArrayList<Double> y,
              int k)

Parameters:

y - The array of dependent values, i.e. averages

k - The start index of subset

Returns:

The mean of dependent values

getSumOfXVariationMultipliedYVariation

public double getSumOfXVariationMultipliedYVariation(ArrayList<Integer> x,
                                            ArrayList<Double> y,
                                            int k,
                                            double xMean,
                                            double yMean)

Parameters:

x - The array of independent values, i.e. observations

y - The array of dependent values, i.e. averages

k - The start index of subset

xMean - The mean of independent values

yMean - The mean of dependent values

Returns:

The sum of multiplication of the variations of independent and dependent values

getSumOfXVariationSquared

public double getSumOfXVariationSquared(ArrayList<Integer> x,
                               int k,
                               double xMean)

Parameters:

x - The array of independent values, i.e. observations

k - The start index of subset

xMean - The mean of independent values

Returns:

The sum of squared variation of independent values

getSumOfRegressionVariationSquared

public double getSumOfRegressionVariationSquared(ArrayList<Integer> x,
                                        int k,
                                        double b0,
                                        double b1,
                                        double yMean)

Parameters:

x - The array of independent values, i.e. observations

k - The start index of subset

b0 - y-intercept of regression line

b1 - The slope of regression line

yMean - The mean of dependent values

Returns:

The sum of squared variation of regression

getSumOfYVariationSquared

public double getSumOfYVariationSquared(ArrayList<Double> y,
                               int k,
                               double yMean)

Parameters:

y - The array of dependent values, i.e. averages

k - The start index of subset

yMean - The mean of dependent values

Returns:

The sum of squared variation of dependent values

getBestSmoothingFactor

public double getBestSmoothingFactor(ArrayList<Integer> x,
                            ArrayList<Double> y,
                            int initialInterval)

Parameters:

x - The array of independent values, i.e. observations

y - The array of dependent values, i.e. averages

initialInterval - The initial interval

Returns:

The best smoothing factor

getAverage

public Double getAverage(ArrayList<Integer> observations,
                ArrayList<Double> averages)

Exponential Moving Average with Dynamic Smoothing Factor

Parameters:

observations - Array that contains peer departures numbers.

averages - Array that contains values of exponential moving averages

interval - The length of interval that is used to calculate smoothing factor

Returns:

The exponential moving average for given observations of peer departures.

getStandardDeviation

public double getStandardDeviation(ArrayList<Integer> range,
                          double currentAverage)

Calculation of Standard Deviation for a set of values

Parameters:

range - A set of values

currentAverage - Average value for the set

Returns:

The standard deviation

getPredictedValue

public Double getPredictedValue(ArrayList<Integer> observations,
                       double currentAverage)

Exponential Moving Average with Dynamic Smoothing Factor plus Deviation

Parameters:

observations - Array that contains recent peer departures numbers.

currentAverage - Current exponential moving average

interval - The length of interval that is used to calculate deviation

Returns:

The exponential moving average plus deviation for a given observation of peer departures. This value is regarded as a prediction value for the next second.

peerInserted

public void peerInserted(PeerAddress peerAddress,
                boolean verified)

Specified by:

peerInserted in interface PeerMapChangeListener

peerRemoved

public void peerRemoved(PeerAddress peerAddress,
               PeerStatatistic storedPeerAddress)

Specified by:

peerRemoved in interface PeerMapChangeListener

peerUpdated

public void peerUpdated(PeerAddress peerAddress,
               PeerStatatistic storedPeerAddress)

Specified by:

peerUpdated in interface PeerMapChangeListener