As network environments with multi-radio access technologies become the norm, dynamically choosing the most appropriate network interface and base station becomes essential for distributing radio resources optimally and fairly, and for allowing each mobile node to take full advantage of its multiaccess capability based on its particular needs and criteria. In these increasingly complex scenarios the importance of handover decisions cannot be underestimated.
This project aims at using simulation to study different handover decision strategies, focusing on the tradeoffs between continuous connectivity, network utilization and performance, and the associated costs. In order to achieve this, we must also investigate and/or develop appropriate and effective simulation tools (for various reasons, the widely-used ns-2 simulation tool is not a good fit for this) : OMNet++ ; developing special-purpose simulation tools using high-level simulation languages ; and so on.In order to evaluate the merits of current and future TCP modifications, reliable models that realistically simulate errors that percolate up to the transport level in hybrid wired and wireless networking environments are needed. We have developed and experimented with an error model framework that is flexible and can recreate many possible scenarios. A thorough analysis of TCP retransmission patterns and dynamics is needed to parameterize and calibrate this error model so as to replicate real-world scenarios.
Some tools already exist for analyzing network-derived traffic traces and extracting TCP-related information from them. This project involves, both : applying these tools to such traces and analyzing the extracted data ; and enhancing the tool set itself in order to achieve improved analytic accuracy and effectiveness.