java多线程模拟M/M/C(马科洛夫队列)

M/M/C(马科洛夫队列)是一个多服务器的排队模型，每个客户到来的时候选择等待队列最短的服务窗口排队等待服务，在商场，银行等服务行业应用广泛，模型如下：

使用服从指数分布的随机数(平均值为lamda)模拟每个客户到达间隔时间，使用服从正态分布的随机数(平均数为mu,标准差为sigma)模拟每个客户在服务窗口接受服务时间，给定服务窗口数，最大客户数，以及lamda，mu和sigma之后，程序进行模拟运算之后可以给出实际计算的lamda，mu和sigma，已经平均队列长度，平均等待时间，平均停留时间等等。程序代码如下：

首先，产生服从指数分布和正太分布伪随机数的算法如下(从网上找到的代码)：

import java.util.Random;

/**
 * Generate the specify type random number
 */
public class RandomGenerator {

	/**
	 * Generate normal distribute random number
	 * @param average Expect average
	 * @param sigma Standard devation
	 * @return Normal distribute random number
	 */
	public static double generateNormalDistrRand(double average,
			double sigma) {
		double N = 12;
		double x = 0, temp = N;
		do {
			x = 0;
			for (int i = 0; i < N; i++)
				x = x + (Math.random());
			x = (x - temp / 2) / (Math.sqrt(temp / 12));
			x = average + x * Math.sqrt(sigma);
		} while (x <= 0);
		return x;
	}
	
	/**
	 * Generate negative index distribute random number
	 * @param lamda Expect average 
	 * @return Negative index distribute random number
	 */
	public static double generateNegativeIndexDistrRand(double lamda) {
		Random random = new Random();
		return -Math.log(random.nextDouble()) / lamda;
	}
}

客户模型：

public class Customer {
	private double arrivalTime = 0.0;//客户到达时刻
	private double serviceTime = 0.0;//客户接受服务的时间
	private double departureTime = 0.0;//客户离开时刻=客户到达时刻+客户等待时间+客户接受服务时间
	private int queueLen;

	public double getDepartureTime() {
		return departureTime;
	}

	public void setDepartureTime(double departureTime) {
		this.departureTime = departureTime;
	}

	public double getArrivalTime() {
		return arrivalTime;
	}

	public void setArrivalTime(double arrivalTime) {
		this.arrivalTime = arrivalTime;
	}

	public double getServiceTime() {
		return serviceTime;
	}

	public void setServiceTime(double serviceTime) {
		this.serviceTime = serviceTime;
	}

	@Override
	public String toString() {//客户停留时间=客户离开时刻-客户到达时刻，客户等待时间=客户离开时刻-客户到达时刻-客户接受服务时间
		return "Customer stay time = " + (departureTime - arrivalTime)
				+ ", and wait time = "
				+ (departureTime - arrivalTime - serviceTime);
	}

	public void setQueueLen(int size) {
		this.queueLen = size;

	}

	public int getQueueLen() {
		return this.queueLen;
	}
}

使用多线程模拟服务窗口：

public class Server implements Runnable {
	private int serverId;
	private volatile boolean isAlive = true;
	private Deque<Customer> customerQueue = new LinkedBlockingDeque<Customer>();//每个服务窗口都有一个客户等待队列

	private int custoemrServed = 0;
	private