Guaranteed service management in mobile wireless Internet

Abstract

The ubiquity of the mobile wireless Internet has led to the increased use of reatime applications. Real-time applications are network resources hungry and propagation delay sensitive. These applications need service guarantees if they are to be acommodated on the mobile wireless Internet. It is, therefore, necessary to maintain the required QoS of these applications in the presence of user mobility with the use of resource reservation mechanism. This dissertation presents development of a resource management mechanism to allocate and guarantee the availability of network resources in the mobile wireless Internet.

A model tagged "Dynamic Mobile Resource Reservation Protocol (DMRSVP)" was proposed. DMRSVP comprises of the following: resource resenaiion, bandwidth allocation, mobility management, and next cell prediction components. The sectored cell approach next cell prediction scheme was modified by adding a resource reservation module. The resource reservation of DMRSVP is made up of a dynamic comparison algorithm for reserving resources in an exactly cell in which a mobile node will visit with open connection. The proposed model utilises HMlPv6 for managing mobility within a domain and the cooperation of hierarchies was adopted to manage mobilyt between domains. An algorithm tagged "connection swapping" was developed for DMRSVP to reduce both call blocking and call dropping probabilities. The simulation of DMRSVP was carried out using an object oriented programming environment, VB.net. The performance of DMRSVP was compared with the existing schemes (MRSVP and HMRSVP) through simulation. The network load, delay, call blocking probability (CBP) and call dropping probability (CDP) were used as the simulation parameters.

The results of the simulalons show that bandwidth utilisation of the three schemes between 0 and 2 seconds (sec) were the same, but after 10 sec, MRSVP utilised 160Kbps of bandwidth, while HMRSVP and DMRSVP utilised 138Kbps and 102Kbps of bandwidth respectively. This verifies thaDMRSVP does not over utilise bandwidth. The proposed model (DMRSVP) blocked 5 reservation requests, while MRS VP and HMRSVP blocked 36 and 24 reservation requests respectively within 10 seconds. The signalling overload of the three protocols shows that MRVP sent 225; HMRSVP sent 151, while the proposed DMRSVP sent 25 control messages over the period of 10 minutes. This implies that the proposed model does not overload the network with a control messages; this allows for more user data packets to be transmitted. The round trip delay of reservation set up of MRSVP was 45 sec; HMRSVP was 31 sec; while it took only 19 sec for DMRSVP to set up the reservation. CBP and CDP of the proposed model were measured. The simulation results show that DMRSVP has CBP of .02 and CDP of 0.05. These are lower values compared to what is obtained in the literature for other schemes.

Based on the results of the simulation, it was concluded that DMRSVP is more scalable and efficient than the classical MRSVP and HMRSVP. This isdue to the fact that DMRSVP takes into account mobility management but MRSVP does not. Again DMRSVP was integrated with the next prediction algorithm to predict the cell(s) in which a mobile node might visit. This makes DMRSVP not to over reserve the limited network resource.