com.net2plan.libraries

Class NetworkPerformanceMetrics

java.lang.Object

com.net2plan.libraries.NetworkPerformanceMetrics

public class NetworkPerformanceMetrics
extends Object

Class providing static methods to compute several performance metrics using classical formulae (i.e. Erlang-B or Kaufman-Roberts recursion for call blocking probability...).

Author:

Pablo Pavon-Marino, Jose-Luis Izquierdo-Zaragoza

Method Summary

Modifier and Type	Method and Description
static double	alphaUtility(DoubleMatrix1D vals, double alphaFactor)
static List<String>	checkNetworkState(NetPlan netPlan, Map<String,String> net2planParameters) Checks the current network state and, maybe, returns some warnings about link oversubscription, demand blocking, and so on.
static double	erlangBLossProbability(int numberOfServers, double load) Returns the probability of call blocking in a M/M/n/n queue system.
static int	inverseErlangB(double gradeOfService, double load) Returns the number of servers (i.e.
static double	jainFairnessIndex(DoubleMatrix1D vals)
static double[]	kaufmanRobertsRecursion(int u_e, double[] h_p, int[] s_p) Computes the Kaufman-Roberts recursion for a multi-rate loss model system.

Methods inherited from class java.lang.Object

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

Method Detail

alphaUtility

public static double alphaUtility(DoubleMatrix1D vals,
                                  double alphaFactor)

checkNetworkState

public static List<String> checkNetworkState(NetPlan netPlan,
                                             Map<String,String> net2planParameters)

Checks the current network state and, maybe, returns some warnings about link oversubscription, demand blocking, and so on.

Important: NetPlan objects are always valid and consistent from a technology-agnostic point of view

Parameters:

netPlan - Input network design

net2planParameters - A key-value map with Net2Plan-wide configuration options

Returns:

List of warnings (empty if none)

erlangBLossProbability

public static double erlangBLossProbability(int numberOfServers,
                                            double load)

Returns the probability of call blocking in a M/M/n/n queue system.

Reference implementation: [1] S. Qiao, L. Qiao, "A Robust and Efficient Algorithm for Evaluating Erlang B Formula", Technical Report CAS98-03, McMaster University (Canada), October 1998

Parameters:

numberOfServers - Number of servers (i.e. link capacity in integer units). It must be greater or equal than zero

load - Traffic load (i.e. carried traffic by the link). It must be greater or equal than zero

Returns:

Call blocking probability

inverseErlangB

public static int inverseErlangB(double gradeOfService,
                                 double load)

Returns the number of servers (i.e. link capacity) to achieve a given grade of service (i.e. call blocking probability) under a given load in a M/M/n/n queue system.

Reference implementation: [1] S. Qiao, L. Qiao, "A Robust and Efficient Algorithm for Evaluating Erlang B Formula", Technical Report CAS98-03, McMaster University (Canada), October 1998

Parameters:

gradeOfService - Grade of service (i.e. call blocking probability). It must be greater or equal than zero

load - Traffic load (i.e. carried traffic by the link). It must be greater or equal than zero

Returns:

Number of servers

jainFairnessIndex

public static double jainFairnessIndex(DoubleMatrix1D vals)

kaufmanRobertsRecursion

public static double[] kaufmanRobertsRecursion(int u_e,
                                               double[] h_p,
                                               int[] s_p)

Computes the Kaufman-Roberts recursion for a multi-rate loss model system.

Parameters:

u_e - Link capacity (in integer units). It must be greater or equal than zero

h_p - Traffic volume vector. Each element is referred to a connection of type p and must be greater than zero

s_p - Capacity units occupied in the link by each accepted connection of type p. Each element must be greater or equal than one

Returns:

Vector of connection blocking probability. Each element is referred to a connection of type p