Lossy Space-Time Filters: An Application Layer for Storing and Querying Spatiotemporal Data

Location is essential in today’s mobile applications, and these applications need system support for efficient storage and retrieval of location information. Spatiotemporal data storage has received significant attention, most notably in spatiotemporal and moving object databases. However, a confluence of systems challenges (driven by constraints on bandwidth, latency, and interactivity) and societal challenges (motivated by desires for privacy) encourage alternatives to massive centralized indexes; on-loading responsibility for location awareness is becoming increasingly popular. We introduce the Lossy Space-Time Filter (LST-Filter), an application layer that enhances spatial data structures in support of this trend. Specifically, the LST-Filter allows a mobile device to assume complete control of storing and sharing the device’s acquired spatiotemporal data—no location data is sent to any centralized index. An LST-Filter is tunable by the user or application to trade storage and computation resources for quality of location information. We show how the layer’s API is tailored to efficiently respond to commonly used mobile application queries (e.g., queries about coverage and trajectories). Using real-world mobility trace data, we benchmark the tradeoffs of LST-Filters and show that, in comparison to naı̈ve on-device location storage mechanisms, the LST-Filter can achieve drastic speedups for these common queries.