Mobility-awareness in complex event processing systems

Abstract

The proliferation and vast deployment of mobile devices and sensors over the last couple of years enables a huge number of Mobile Situation Awareness (MSA) applications. These applications need to react in near real-time to situations in the environment of mobile objects like vehicles, pedestrians, or cargo. To this end, Complex Event Processing (CEP) is becoming increasingly important as it allows to scalably detect situations “on-the-fly” by continously processing distributed sensor data streams. Furthermore, recent trends in communication networks promise high real-time conformance to CEP systems by processing sensor data streams on distributed computing resources at the edge of the network, where low network latencies can be achieved. Yet, supporting MSA applications with a CEP middleware that utilizes distributed computing resources proves to be challenging due to the dynamics of mobile devices and sensors. In particular, situations need to be efficiently, scalably, and consistently detected with respect to ever-changing sensors in the environment of a mobile object. Moreover, the computing resources that provide low latencies change with the access points of mobile devices and sensors. The goal of this thesis is to provide concepts and algorithms to i) continuously detect situations that recently occurred close to a mobile object, ii) support bandwidth and computational efficient detections of such situations on distributed computing resources, and iii) support consistent, low latency, and high quality detections of such situations. To this end, we introduce the distributed Mobile CEP (MCEP) system which automatically adapts the processing of sensor data streams according to a mobile object’s location. MCEP provides an expressive, location-aware query model for situations that recently occurred at a location close to a mobile object. MCEP significantly reduces latency, bandwidth, and processing overhead by providing on-demand and opportunistic adaptation algorithms to dynamically assign event streams to queries of the MCEP system. Moreover, MCEP incorporates algorithms to adapt the deployment of MCEP queries in a network of computing resources. This way, MCEP supports latency-sensitive, large-scale deployments of MSA applications and ensures a low network utilization while mobile objects change their access points to the system. MCEP also provides methods to increase the scalability in terms of deployed MCEP queries by reusing event streams and computations for detecting common situations for several mobile objects.