Article Preview
Top1. Introduction
The dispersion of networks and the internet throughout the world increases rapidly. The internet components are made up of a set of nodes, routers, switches, computer devices, and others (Abbasov & Korukoglu, 2009; Brandauer et al., 2001; Floyd, 2000; Stanojevic et al., 2006; Woodward, 1994). Each of these components has a specific role and function in accordance to specified protocols and constrained policies. The dominant function of the router is to bridge the connection between computer networks and ensure packets transmission to their destination (Choudhary & Goyal, 2021). In addition, the router prevents any unauthorized access to the networks (Derakhshanfard & Soltani, 2021). Therefore, for the data to be spread between the computer networks, it has to be stored in the router buffer, and processed. Then, it will be forwarded as chunks to its destination via the router using First-In First-Out (FIFO) method as shown in Figure 1 (Abbasov & Korukoglu, 2009; Goyal, 2020; Woodward, 1994).
Congestion phenomenon is the cause behind packet loss and delay in the computer networks (Abbasov & Korukoglu, 2009; Goyal et al., 2016; Kumar et al., 2020; Sarker et al., 2021; Stanojevic et al., 2006). When the rate of packets arrival increases overwhelmingly exceeding departure rate, the router buffer will be stacked up with data packets since the router capacity is limited. (Zhang et al., 2014). In addition, when the buffer starts to overflow, the packet dropping ratio will increase as there is insufficient space to host the arriving packets.
When the queue size length increases, the probability of packet arrival also increases which causes a failure in network status analysis and congestion detection (Kazmi et al., 2019). Therefore, several Active Queue Management methods (AQM) have been proposed to facilitate the congestion detection such as drop tail (Stanojevic et al., 2006). To avoid the congestion, the packet transmission rate should be reduced for a period of time then increased as the slowing rate time was over. This leads to congestion reoccurrence as the senders increase the rate congruently which is known as global synchronization phenomenon. The size of the queue is limited based on routers limitation. The senders will reduce their transmissions for a limited time while packet loss occurred based on packet queue. After a time delay, the senders start again resulting in overflow and more packets getting dropped. The drop tail method drops the packets stochastically based on dropping probability in order to prevent such situation from happening (Baklizi, 2020; Stanojevic et al., 2006).
There are some congestion indicators that should be measured to validate the performance of the network and router buffer such as q which denotes the number of packets present in the router buffer (Dutta, 2020; Gupta & Goyal, n.d.). The dropping probability performance measures the value in the range of [0-1]. In case of the calculated dropping probability being equal to zero, no dropping occurs as the arriving packet is accommodated upon arrival. Yet, if the calculated dropping probability is equal to one, the arriving packet is discarded. Lastly, if the calculated dropping probability is between zero and one, the arriving packet is either accommodated or discarded, stochastically, to ensure that no congestion occurs and simultaneously prevents the global synchronization phenomenon (Ziaeemehr et al., 2020).