Vol.:(0123456789) 1 3 Int. j. inf. tecnol. https://doi.org/10.1007/s41870-023-01201-1 ORIGINAL RESEARCH Modifed dropping‑random early detection (MD‑RED): a modifed algorithm for controlling network congestion Samuel Hassan 1  · Adewole Rufai 2   Received: 20 September 2022 / Accepted: 21 February 2023 © The Author(s), under exclusive licence to Bharati Vidyapeeth’s Institute of Computer Applications and Management 2023 Abstract The quality of real-time network services is greatly dependent on the infuential role of delay. This prob- lem of high end-to-end delay is due in part to the presence of a congested network. Active queue management (AQM) algorithm accomplished at the router has since emerged as a robust area of research with a myriad of published works. One algorithm with a high hype in the research area is the random early detection (RED) proposed well over 2 decades ago. Previous enhancements to RED are unable to convinc- ingly reduce the average queue size and improve network delay. To address these challenges that RED’s one drop function is unable to meet, in this paper, wherefore, a new amendment to RED is described, namely modifed dropping- RED (MD-RED). Contrastingly, in MD-RED, a linear drop function is tailored for light and moderate loads to establish a proportional growth in packet dropping while an exponen- tial drop function is tailored for high trafc load to guaran- tee a more rapid increase in packet dropping. Reports from experiments conducted in ns-3 open-source simulation tool confrms that MD-RED suitably controls network congestion and ofers an appreciable improvement in network delay. Keywords AQM · Congestion control · MD-RED · Communication networks · Simulation 1 Introduction To start with, Internet trafc increases with high magnitude each day, due to ever-growing active number of users utiliz- ing a myriad of applications that runs on the Internet [1–4]. This phenomenal exponential growth in connectivity poses the critical problem of increased network congestion [3, 5–7]. In a more deeper sense, network congestion happens when the aggregate trafc loads becomes larger than the available processing capacity of the router [7–9]. The problem of network congestion could lead to perfor- mance degradation in the Quality of Service or QoS ofered in network and communication systems if not monitored and efectively controlled [10, 11]. Furthermore, network life- time can be enhanced by the use of an efective congestion control algorithm [12, 13]. Congestion control algorithms are broadly categorized into two perspectives: source-based and network-assisted [3]. The former is achieved through the use of transport layer protocol—Transport Control Protocol, better known as TCP. Whereas, the later is carried out through the involvement of intermediate devices (in other words, the router). Network routers perform an indispensable role of provid- ing queue management aside its usual role of packet rout- ing [9, 14, 15]. Generally, queue management algorithms are implemented at the routers and addresses congestion by means of dropping or accommodating packets whenever its necessary at the bufer [9, 16]. Queue management algorithms are of two forms: pas- sive and active. The principal concept of active queue management, better known as AQM is to detect congestion at an introductory stage and sends a congestion notifcation (by-means-of dropped packets) to transmitting end-nodes so that they can respond by minimizing the rate of packet generation [17–19]. By far, AQM algorithm ofers a vast * Samuel Hassan samuel.hassan@oouagoiwoye.edu.ng Adewole Rufai arufai@unilag.edu.ng 1 Department of Mathematical Sciences, Olabisi Onabanjo University, Ago-Iwoye, Ogun, Nigeria 2 Department of Computer Sciences, University of Lagos, Akoka, Lagos, Nigeria