The application of artifical intelligence for network management

The B-ISDN networks enjoy a wide bandwidth that is compatible and enough to conveythe multimedia traffic between the individual networks. These networks enjoy high interactivecapability and intelligent management tools that suit the new multimedia computer applications.These abilities are the infra structures for the ATM networks. The ATM networks use the VirtualChannels (VC) and Virtual Path (VP) for the routing policies. Most of the researches [96, 136] usethe linear search tools to seek for an available Virtual Channel (VC). The linear search consumesan average seeking time proportional to (n+ 1 )/2 , where ”n” is the number of channels. Theseeking time can be reduced using the bit mask tools which search in a parallel manner and canselect the VC in one slot. This improves the seeking time in the network by a ratio equals (n+ 1 )/2.There still other set of problems need solutions such as the traffic control algorithms, routingprotacols and Bandwidth Allocation (BA) tools. There is also some problems regarding the ATMnetworks such as the loss ratio in the burst traffic, cell transmission delay and the delay variationof the individual signals [94]. This thesis applies some Artificial Intelligence (AI) tools to enhancethe performance of computer networks. The AI tools are applied for the networkmanagement during this thesis in the following forms:Collecting large knowledge base either periodically or asynchronousely to estimate thermance metrics of the ATM networks.11- Using ALARMS as learning tools to notify the connected servers to use the alternative oremergency solutions. This resilient the network performance.Using the prioritizing protocols at each Network Termination (NT) to filter the crossing traffic so asit remains less than the bandwidth of the server. This results congestion free conditions. This, is modified using a heuristic function that increases the static priority regarding the time, ’on of the cell and the service availability.dynamic priority = Availability * time variation of the cell+ static priority of the class.Using bandwidth allocation algorithm to improve the effective bandwidth of the network.The review of literature during this thesis is explored within the first four chapters. The nextchapters include the new models, theorems and the simulation results for the thesisbibutlon,Chapter .-1: ARTIFICIAL INTELLIGENCE, TOOLS and ALGORITHMS.chapter explored the knowledge representation and reasoning tools that are used for ArtificialDlgence Systems (AIS). The AIS use heuristic approaches that are more efficient thanentional algorithmic approaches .Chapter -2: COMPUTER NETWORKS TECHNOLOGIES