Time series data from sensor devices are increasingly stored in log data structures across the cloud and mobile devices. Currently, log data is accessed as chunks of fixed size, which enhances performance by prefetching of data. However, in applications such as remote monitoring of patients using mobile devices, data requirement of end users varies significantly depending upon their roles. The fixed chunking approach would lead to unnecessary data download due to the dynamic variability of data access. Also, the requests are more often than not based on fixed time chunks that do not necessarily translate to fixed data size. To overcome this challenge, we present a dynamic log chunking mechanism based on reader access pattern and domain specific data characteristics. The application of this method in the area of remote patient monitoring in bandwidth starved rural areas is shown to result in bandwidth and cost savings of 14% without affecting the prefetch performance.
Rahul K Pathinarupothi, Bithin Alangot, and Rangan, V., “Context Aware Dynamic Log Chunking for Mobile Healthcare Applications: Poster”, in Proceedings of the 17th ACM International Symposium on Mobile Ad Hoc Networking and Computing, New York, NY, USA, 2016.