Modeling frequently accessed wireless data with weak consistency

Abstract

To reduce the response times of wireless data access in a mobile network, caches are utilized in wireless handheld devices. If the original data entry has been updated, the cached data in the handheld device becomes stale. Thus, a mechanism is required to predict when the cached copy will expire. This paper studies a weakly consistent data access mechanism that computes the time-to-live (TTL) interval to predict the expiration time. We propose an analytic model to investigate this TTL-based algorithm for frequently accessed data. The analytic model is validated against simulation experiments. Our study quantitatively indicates how the TTL-based algorithm reduces the wireless communication cost by increasing the probability of stale accesses. Depending on the requirements of the application, appropriate parameter values can be selected based on the guidelines provided in this paper.