Search

Spatio-temporal events often describe the movements of an object in terms of space, time, and potential other attributes. Significant knowledge can be inferred by analysing them, either individually or atomically in form of trajectories. The trajectories can abstract additional properties and lead to deeper value. Moreover, external contextual information can be attributed to them to change their structure and lead to different perspectives. Because of this potentially valuable knowledge, nowadays indoor and outdoor tracking devices are used everywhere; generating countless events instantaneously. However, the extraction of knowledge from such heterogeneous, massive data is not a trivial task. In other terms, there is a need for a sophisticated system that is efficient in terms of distributed computing, failure handling, responsiveness, and abstraction. To answer this need, our study incorporates a fully fledged, reactive system for big trajectory data management. The system is unique of its kind because it is actor-based and features responsiveness, resiliency, and elasticity. Furthermore, our system is implemented using Scala; hence, we have the expressive power of both the Object-Oriented (OO) and Functional Programming (FP) paradigms. Allowing us to reach a higher level of abstraction to be able to process any trajectory type. The scope of this paper is to detail our system and discuss elasticity, routing strategies, load balancing, and our proper fault-tolerance mechanism. To fulfill this study, we consider the Geolife project’s GPS trajectory dataset.

Nowadays, distributed systems have become requisite to process and analyse the large amount of generated data. In particular, spatio-temporal data has known an exponential growth in the last years. This could be explained by the proliferation of indoor and outdoor tracking devices and the value of the knowledge that can be extracted from their analysis. The data describe a moving objects’ behavior in space and time. Once these coordinates are assembled they form a raw trajectory. The knowledge extracted by analysing the set of trajectories is very valuable; they can be even mapped to other contextual data to add value. Preprocessing is an essential step in the mining process to filter unnecessary records and clean the noise. Furthermore, reactive systems offer the possibility to process massive data in a non-blocking, resilient, responsive and elastic manner. Despite this interest, to the best of our knowledge no works have been conducted into offering a reactive system to preprocess massive trajectories. The aim of this study is to fill this gap by proposing a reactive system based on distributed actors to manage big trajectory data. Our system is deployed with the Play Framework, Akka, MongoDB, AngularJS and D3.js. Initially, the system can load batches of trajectories stored in HDFS in a distributed manner. The scope of our study is to provide an overview of the system, to study the impact of the increase in computing resources over the scalability, and to provide the optimal node configuration. The results indicate a higher scalability of the system, and the evaluation is conducted by considering the Geolife project’s GPS trajectory dataset.

Reactive Systems, especially in the Big Data Era, are undergoing a revolution. The increasing interest is due to their responsiveness, elasticity, resiliency and message-driven trait. In particular, a reactive message-driven system has one core feature asynchronous communication assuring a non-blocking behavior. A particular case of such systems is the use of actors to assure encapsulation and support fast decisions, failure handling and progess. The Actor Model meets the requirements for buiding a highly conccurent, scalable, distributed system, and Akka's tools provide the means to build it. However, to the best of our knowledge the effect of intra-actor asynchronous message processing on the routing strategies and the load balancing issue has not been addressed in the literature yet. The aim of this study is to analyse the behavior of the routers and routees over the existing routing strategies when each routee processes its messages asynchronously. Moreover, we study how the load is balanced among the routees and propose our own routing strategy that can reduce the processing time overhead by considering the complexity of the messages. For that, we build a distributed system to offline process massive trajectories, initially stored in HDFS, with Akka, the Play Framework, and MongoDB. Specifically, we leverage the Geolife projectâArZs GPS trajectory dataset.

The proliferation of indoor and outdoor tracking devices has led to a vast amount of spatial data. Each object can be described by several trajectories that, once analysed, can yield to significant knowledge. In particular, pattern analysis by clustering generic trajectories can give insight into objects sharing the same patterns. Still, sequential clustering approaches fail to handle large volumes of data. Hence, the necessity of distributed systems to be able to infer knowledge in a trivial time interval. In this paper, we detail an efficient, scalable and distributed execution pipeline for clustering raw trajectories. The clustering is achieved via a fuzzy similarity relation obtained by the transitive closure of a proximity relation. Moreover, the pipeline is integrated in Spark, implemented in Scala and leverages the Core and Graphx libraries making use of Resilient Distributed Datasets (RDD) and graph processing. Furthermore, a new simple, but very efficient, partitioning logic has been deployed in Spark and integrated into the execution process. The objective behind this logic is to equally distribute the load among all executors by considering the complexity of the data. In particular, resolving the load balancing issue has reduced the conventional execution time in an important manner. Evaluation and performance of the whole distributed process has been analysed by handling the Geolife project&rsquo, s GPS trajectory dataset.