UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Automated network selection for service delivery across all-IP heterogeneous wireless systems Bari, Farooq

Abstract

A crucial step in achieving service delivery in heterogeneous all-IP wireless systems is the selection of an appropriate delivery network. Simple network selection mechanisms as they exist today cannot be applied in this new service environment. Because of the revenue and service impacting nature of the problem, it has been considered critical by the communication industry to find a good solution. In our research we have investigated the scope of the problem and addressed both the architectural and algorithmic aspects of the problem. We have proposed an architectural framework that is based on the network assisting the terminal in the decision about network selection. The proposed solution minimizes usage of the wireless links and would work with currently deployed infrastructure. It is scalable, flexible and supports roaming. We developed a two step decision process for a network-assisted network selection mechanism that combines non-compensatory and compensatory multi-attribute decision making (MADM) algorithms. We have proposed enhancements to several MADM algorithms for their application to network selection. In order to deal with ranking abnormalities, we have proposed an improvement to the standard TOPSIS algorithm when it is applied to network selection by adopting an iterative approach. We have identified the need for support of non-monotonic utility for attributes used in network selection decision making and demonstrated that GRA is better suited for achieving this type of optimization objective. We have modified ELECTRE so that it can be used for network selection with attributes exhibiting non-monotonic utility. Our contributions in terms of modification to TOPSIS and ELECTRE and the usage of GRA with non-monotonic utility are not specific to the problem of network selection but these ideas can be used in other problems with similar requirements. We have developed a decision strategy for network selection in the absence of precise input information. The decision process proposed by us uses fuzzy techniques along with data prediction for the network selection decision process in certain cases. We have developed a new function, Confidence Level, and used it as a decision support tool along with the sensitivity of the service/subscription to the missing/imprecise information. A crucial step in achieving service delivery in heterogeneous all-IP wireless systems is the selection of an appropriate delivery network. Simple network selection mechanisms as they exist today cannot be applied in this new service environment. Because of the revenue and service impacting nature of the problem, it has been considered critical by the communication industry to find a good solution. In our research we have investigated the scope of the problem and addressed both the architectural and algorithmic aspects of the problem. We have proposed an architectural framework that is based on the network assisting the terminal in the decision about network selection. The proposed solution minimizes usage of the wireless links and would work with currently deployed infrastructure. It is scalable, flexible and supports roaming. We developed a two step decision process for a network-assisted network selection mechanism that combines non-compensatory and compensatory multi-attribute decision making (MADM) algorithms. We have proposed enhancements to several MADM algorithms for their application to network selection. In order to deal with ranking abnormalities, we have proposed an improvement to the standard TOPSIS algorithm when it is applied to network selection by adopting an iterative approach. We have identified the need for support of non-monotonic utility for attributes used in network selection decision making and demonstrated that GRA is better suited for achieving this type of optimization objective. We have modified ELECTRE so that it can be used for network selection with attributes exhibiting non-monotonic utility. Our contributions in terms of modification to TOPSIS and ELECTRE and the usage of GRA with non-monotonic utility are not specific to the problem of network selection but these ideas can be used in other problems with similar requirements. We have developed a decision strategy for network selection in the absence of precise input information. The decision process proposed by us uses fuzzy techniques along with data prediction for the network selection decision process in certain cases. We have developed a new function, Confidence Level, and used it as a decision support tool along with the sensitivity of the service/subscription to the missing/imprecise information.

Item Media

Item Citations and Data

License

For non-commercial purposes only, such as research, private study and education. Additional conditions apply, see Terms of Use https://open.library.ubc.ca/terms_of_use.

Usage Statistics