The fundamental property of Markov chains, called Markov property is that its future development will depend on the past only through its current value. The figure below gives the number of servers, the letters that follow (optional) identify the size of the queue and the size of the queue management solutions source population (if these values are omitted, they are assumed infinite). For example, M/ M/ refers to the base model: Poisson arrivals, exponential service, a single server queue unbounded.

 

M/ D/ K/ K indicates consistent service queue management solutions time, K and K servers seats total (no waiting: model). In a queue, a parameter of primary importance is the utilization rate, traditionally denoted. It is the product of the arrival rate of customers by service time. We can show that 1-ρ gives the queue management display probability of finding the inactive server. It takes (for a queue of infinite capacity) that (stability condition). A standby M/ M/ queue is a system consisting of a line of infinite capacity and one server.

Guests arrive in the system according to a Poisson process with rate knowing that is the inter-arrival time, A (t) is a random variable with exponential distribution with parameter. The service time is distributed according to an exponential distribution with parameter knowing is the queue management display average size of a packet. Despite all the efforts to control the information flow in a network, it may face a problem of congestion. It is therefore to solve the problem without increasing it.

 

Indeed, congestion problems arrive when the nodes of a network saturate their queues and thus lose packets. The congestion is unavoidable statically on a queue management solutions network. It is the result of increased delivery times. If the size of the queues in the nodes increases, the blocks are not taken on time and are retransmitted, which increases the flow. The congestion occurs when the queue management display flow (load) in the network exceeds the resource capacity.

Among the major causes of congestion in the network are: The goal of congestion control is to try to ensure a good quality of service to users. In a congestion problem (load node too high), buffer memory nodes are saturated, new received packets are queue management solutions dropped causing retransmissions; hence the need for congestion control. The Explicit Congestion Notification ECN is an addition to the TCP/ IP stack designed in 1994 and standardized in September 2001 by the IETF as RFC 3168 [3]. It is a mechanism that anticipates congestion on a router.

ECN is only used when the two protagonists (source and destination) report that they are willing to use this extension. With this method, an ECN flag is placed in queue management display the IP header to explicitly inform the sender that the router is. ECN is an alternative to indirect congestion notification made by deleting a package queue management solutions executed by RED. XCP is a congestion control protocol suggested by Mark Hadley and developed by Dina Katabi.