Evaluation of authentication latency using predictive content transfer to support fast handover

Abstract

In modern era, wireless networks and growing number of electronic devices that are Wi-Fi enabled such as smart phones and laptops are becoming popular. Unfortunately in the busy world nowadays, users are always on the move and will experience difficulty in re-connecting to a new Access Point (AP) especially when they are using real-time applications. The problem is due to the handover process to a new AP that takes time which may cause interruption to the applications that are in use. Furthermore, the handover process undergoes several phases including authentication phase which results in high latency to the user. In order to tackle the issue, the user actually needs a smooth handover from one AP to another, especially within an enterprise Wi-Fi network. Many approaches have been taken in consideration to solve the problem of authentication latency. One of them is where the situation can be resolved if the context transfer can occur prior to the handover process to the next AP. In this thesis, we study on designing the predictive algorithm that triggers context transfer in a centralized network. Based on the proposed method, we conduct an experimental analysis on authentication latency by comparing two scenarios; ‘without’ Predictive Context Transfer and ‘with’ Predictive Context Transfer. The success of this predictive algorithm is determined by the time saved during the context transfer, and ideally the mobile user will experience a fast handover, thus causing minimal interruption to the real-time applications that are in use. The amount of improvement is expected to be close to the time required for authentication, since authentication happens prior to handover process in Predictive Context Transfer.