الفهرس 
الغلافOnly 14 pages are availabe for public view
 |  | from | 129 |  |  |
| | | from | 129 | | |
Abstract
Network congestion occurs when the network load is greater than the network capacity. When the network has too much traffic, congestion occurs and the performance decreases sharply. Congestion control adjusts the rate of sending data to reduce the number of packet losses, time delay, and increasing the packet delivery ratio. There are two types of algorithms working on congestion control which are host-centric algorithms and router centric algorithms. The host-centric algorithms work on host side such as slow start, congestion avoidance, fast retransmit, fast recovery TCP Reno, TCP New Reno, SACK, TCP Vegas. The router centric algorithms work on the router side to control the congestion such as RED, Adaptive RED, MRED, BLUE, A-CHOKE, and P-CHOKE. In this thesis, we concentrate on how to make congestion control in a high-speed network that is also known as a long fat network “LFN” or elephant. Context aware monitors the environment of the network and offers options or actions by making use of high bandwidth-delay product “BDP” to reach a high utilization of the network capacity in LFN.
The congestion control algorithms in LFN are either basic or hybrid congestion control algorithms. Researchers made a comparison between algorithms to get which algorithm has the best performance to control the congestion in LFN. The result of their comparisons proves that Yet Another High-speed TCP algorithm (YEAH-TCP) has the highest performance among the other algorithms but it still suffers from packet losses. Hence this thesis provides a new hybrid approach called MP-YEAH algorithm ”Multipath-YEAH” that utilizes multipath routing techniques. MP-YEAH improves the performance of YEAH with high throughput, reducing delay with 7.3%, reducing energy consumption with 4.1%, and increasing packet delivery ratio with 24%. Deep packet inspection (DPI) works on the whole packet and makes a deep analysis for the packets in the traffic. DPI deeply deliberates context-awareness in packets traffic. In this thesis, a Modified DPI is proposed by considering the use of “MPYEAH” to be more powerful in congestion control in LFN and improves the weakness of the original DPI.
The Modified DPI with MPYEAH gives us an improvement in two sides of view. The first one, Modified DPI with MPYEAH achieves an efficient and powerful congestion control system higher than the original DPI in performance and makes the network more intelligent. The second one, Modified DPI with MPYEAH improves the performance of the MPYEAH algorithm by decreasing the packet dropping, increasing throughput with 25%, and reducing the jitter delay with 31%.
.