Project Home
PDF access

Authors

Abdelrahman Salah, Nayra Mawardy, Shams Abdelhamid, Zeina Ayman Supervised by: Dr. Osama Talaat, Eng. Lobna Mostafa, Eng. Heba Bakr Seddik Osman.

Publishing Date

May 16, 2022

Abstract

With the growing communication networks and the increased number of applications, services and users, it’s challenging to produce a network that efficiently adapts to that growth with flexible and dynamic allocation. In this paper, our main goal is to enhance users’ QoS (Quality Of Service) by considering their behavior and thus optimizing their devices while browsing. We want to do that using SDN (Software-Defined Network) which can determine the priority needed for each user and then allocate the bandwidth accordingly using DBA (Dynamic Bandwidth Allocation) to control the network traffic. Priority assignment would be constructed by using DPI(Deep Packet Inspection) to inspect the packet’s content and see user behavior. It would also predict future usage for the device and automatically assign it a bandwidth. Using priority assignment per device would guarantee user satisfaction.

1.1 Background

Bandwidth allocation has been developed and used for many years with many variant techniques to guarantee QoS ( quality of service). More algorithms and techniques are developed to enhance its performance. In addition, scheduling algorithms were conducted to be used in ordering the packet buffers. They depend on a variable that differs from an algorithm to another such as weight, priority values, or packet arrival time and many other factors including PQ (Prioriyt Queue), WFQ(Weighted Fair Queuing), FIFO(First In First Out) [3], CBWFQ (Class Based Weighted Fair Queue) and LLQ (Low Latency Queuing) [4]. Bandwidth allocation in indoor environments depends on scheduling algorithms implemented in routers that organize the order of packets waiting to be served in the buffers queue. By controlling how these packets are ordered in the queue and served by the router bandwidth allocation can be affected to the favor of optimizing the Quality of Service (QoS). Before the concept of SDN was introduced, many networks relied on using ordinary conventional methods. The traditional method is very complex and different to configure especially because it is hardwarebased. It is fixed within network devices like switches and routers. One of the negative effects of using traditional networks is its limitations. It poses a problem where every component is integrated with each other such as the control plan (decides how to handle the traffic) and data plane (forwards traffic based on the decisions of the control plane) [5] as shown in Figure 1. It can suffer from change and flexibility. However, using SDN architecture is software-based. Because of that, it is more flexible, providing greater control. It also separates the control plane and the data plan as shown in Figure 2 which makes it easier to create abstractions like simplifying, specification and forwarding.

1.2 Motivation

How to improve bandwidth utilization has always been a problem worth continuous study and research. The problem is becoming a real concern nowadays due to the increase of network services. Currently, users’ activities and behaviors are not taken into consideration by internet service providers (ISP) when performing bandwidth allocation. Instead, ISP focuses on the service or application used by the user to determine the suitable bandwidth to be allocated and ignores the user characteristics factor which may not achieve the highest level of users’ satisfaction. There are various proposed solutions for the problem, one of them was proposed by H. Pham-Thi, H. Hoang-Van et al (2014) [5]. They pointed out the fact that allocating the same amount of resources to all users might not meet the users’ expectations. Thus, they proposed classifying the users into 3 groups depending on their network demands: relaxed, short busy, and long busy users resulting in a fair optimized allocation and management of resources. Another solution is proposed by N. E. Frigui et al [6]. They had 2 approaches to optimize dynamic bandwidth allocation (DBA), one was to modify the way in which DBA works on the algorithm itself. However, due to equipment supplier dependency, the network operator does not have full control over such a mechanism. The other and more suitable approach they proposed is to manage the external parameters of the DBA without modifying the algorithm itself. Such parameters can be adjusted dynamically to estimate the users’ traffic patterns and behaviors linked to daily activities. Pham-Thi, Huong et al [7] proposed taking users’ satisfaction and physiological factors into more consideration while allocating the bandwidth. Accordingly, they classified users into 3 groups: relaxed, neutral, and not relaxed. They used the mean opinion score (MOS) to calculate the satisfaction levels. Although there are various proposed solutions to allocate bandwidth based on user characteristics, using SDN in our proposed solution to control and configure the network will ease the process and provide more flexibility.SDN will be used to assign the priority to each user and optimize the bandwidth allocation based on user behaviors.

1.3 Problem Statement

Bandwidth allocation in indoor places had many approaches and techniques that vary in their flexibility and efficiency. One of their approaches is that the ISP assigns a period of the bundle time limit to test and analyze the bandwidth usage by the device connected in this period. During that period, bandwidth is allocated to work with maximum effort blindly, then after analyzing, bandwidth allocation becomes more relative to the quote consumption and handling the network abuse. Bandwidth allocation focuses on the end device or the application as its main factor to allocate the bandwidth or priority and that ignores the human behavior that is actually behind the usage of the device or the APP. Human daily routine activities and behaviors for every range of time (time frame) will be the main factor to set the optimum priority for the device that represents the user during a specific time frame, so the user behavior will be studied and used as a main factor to decide the main bandwidth allocation needed by a specific device in a specific time frame. The other obstacle is the hardness in modifying the logic of routers to customize it to a developed purpose, but by using a modern technology called SDN (Software Defined Network), that inflexible logic implemented in routers will be flexible and programmable as the proccesing happens in a server device called SDN RYU controller