An adaptive approach to indexing pervasive data

Publisher Website
Google Scholar
Abstract
In a pervasive computing world data will be scattered among millions of devices and finding the right data will be a fundamental problem. Several proposed service discovery architectures can assist applications searching for data within local boundaries but there is currently no support for global access to data. We introduce an application-level protocol VIA* for building self-organizing, distributed, hierarchical data indices that adapt to dynamic query workloads. These indices efficiently route queries to relevant devices and reduce the overall workload of the system. Adapting to the query workload, VIA* uses a “query impedance” metric to approximate the optimal hierarchy for processing the expected query workload. Distributed, “logical” nodes in the interior of the hierarchy collect information about query impedance and forward this information to “data carrying” leaf nodes that react to improve the topology of the hierarchy. We present some findings from our workload testbed that demonstrate the performance and scalability characteristics of our approach and outline our research agenda

This publication has 3 references indexed in Scilit: