Showing papers by "Yannis Theodoridis published in 2020"

Semantic-aware aircraft trajectory prediction using flight plans

Abstract: Aircraft trajectory prediction (TP) is a challenging and inherently data-driven time-series modeling problem Adding annotation or enrichment parameters further increases the search space complexity, especially when ‘blind’ optimization algorithms are employed In this paper, flight plans, localized weather and aircraft properties are introduced as trajectory annotations that enable modeling in a space higher than the typical 4-D spatio-temporal A multi-stage hybrid approach is employed for a new variation of the core TP task, the so-called Future Semantic Trajectory Prediction, including clustering the enriched trajectory data using a semantic-aware similarity function as distance metric Subsequently, a separate predictive model is trained for each cluster, using a nonuniform graph-based grid that is formed by the waypoints of each flight plan In practice, flight plans constitute a constrained-based training of each predictive model, one for each waypoint, independently The proposed method is formulated and experimentally validated with real aviation dataset (flight plans and IFS radar tracks) and localized weather data for a 1-month time frame of flights in the Spanish airspace Various types of predictive models are tested, including hidden Markov model (HMM), linear regressors, regression trees and feed-forward neural networks The results show very narrow confidence intervals for the per-waypoint TP errors in HMM, while the more efficient linear and nonlinear regressors exhibit 3-D spatial accuracy much lower than the current state of the art, up to a factor of five compared to ‘blind’ TP for complete flights, in the order of 2–3 km compared to the actual flight routes

Distributed Subtrajectory Join on Massive Datasets

Abstract: Joining trajectory datasets is a significant operation in mobility data analytics and the cornerstone of various methods that aim to extract knowledge out of them. In the era of Big Data, the production of mobility data has become massive and, consequently, performing such an operation in a centralized way is not feasible. In this article, we address the problem of Distributed Subtrajectory Join processing by utilizing the MapReduce programming model. Compared to traditional trajectory join queries, this problem is even more challenging since the goal is to retrieve all the “maximal” portions of trajectories that are “similar.” We propose three solutions: (i) a well-designed basic solution, coined DTJb; (ii) a solution that uses a preprocessing step that repartitions the data, labeled DTJr; and (iii) a solution that, additionally, employs an indexing scheme, named DTJi. In our experimental study, we utilize a 56GB dataset of real trajectories from the maritime domain, which, to the best of our knowledge, is the largest real dataset used for experimentation in the literature of trajectory data management. The results show that DTJi performs up to 16× faster compared with DTJb, 10× faster than DTJr, and 3× faster than the closest related state-of-the-art algorithm.

Citywide Traffic Analysis Based on the Combination of Visual and Analytic Approaches

Abstract: A method for citywide traffic analysis is introduced based on the combination of visual and analytical approaches. Large volumes of GPS data collected from urban vehicles are utilized. In the method, a traffic condition map is constructed, composed of five different layers featuring traffic conditions, road linkage, travel patterns, congestion zones, and traffic flows, respectively. Based on the map, specific transport situations surrounding the congested areas are examined and ways of reducing congestion are suggested. The method is evaluated in the aggregated metropolitan area of Athens and Piraeus in Greece, and the potential and the effectiveness of this technique in analysing traffic are demonstrated. With more and more urban vehicles being equipped with GPS devices, the method can be easily transferable to other regions, paving the way for the adoption of the approach for an up-to-date, spatial-temporal sensitive, visual and analytic method for traffic monitoring that supports the establishment of a more sustainable urban transportation system.

Sea Area Monitoring and Analysis of Fishing Vessels Activity: The i4sea Big Data Platform

Abstract: The i4sea research project provides effective and efficient big data integration, processing and analysis technologies to deliver both real-time and historical operational snapshots of fishing vessels activity in national sea areas. This paper presents the architecture of the i4sea big data platform for sea area monitoring and analysis of fishing vessels activity and demonstrates the operation of some use-case pilot scenarios.

Future Location and Trajectory Prediction

Abstract: This chapter presents modern approaches and frameworks for predicting trajectories with detailed descriptions of three main research pillars The first pillar is the problem formulation regarding two complementary tasks, namely the Future Location Prediction (FLP) and the Trajectory Prediction (TP) The second pillar tackles the issue of effectiveness, efficiency, and scalability for the corresponding predictive analytics models for big fleets of moving objects Finally, the third pillar takes into account historical patterns and semantically rich contextual information, so as to improve the prediction accuracy, especially for long-term time windows The overall assessment of these methods shows that the suite of FLP and TP algorithms developed addresses all the major prediction challenges regarding mobility patterns in terms of points or trajectories, respectively It is expected that these modeling approaches can be transferred to other domains of similar challenges and with similar success

Trajectory Detection and Summarization over Surveillance Data Streams

Abstract: In this chapter, we present Synopses Generator, a stream-based processing framework that can provide online summarized representations of trajectories specifically for sailing vessels and flying aircraft Assuming that surveillance data monitoring their locations over a large geographical area is available in a streaming fashion, this novel methodology drops any predictable positions (along trajectory segments of “normal” motion characteristics) with minimal loss in accuracy Effectively, it can keep only those positions conveying salient mobility events (annotated as stop, change in speed, heading, or altitude, etc), identified when the mobility pattern of a given vessel or aircraft changes significantly Moreover, this framework specifies parametrized conditions for detecting such mobility features, as well as suitable heuristics that can eliminate inherent noise and can provide succinct trajectory synopses in one pass over the incoming streaming positions A prototype implementation on top of Apache Flink and Kafka has been set up in modern cluster infrastructures to enable parallelization of the trajectory summarization process against such big mobility data A comprehensive experimental evaluation has been conducted against various surveillance data in the maritime and aviation domain, and offers concrete evidence of its timeliness, scalability, and compression efficiency, with tolerable concessions to the quality of resulting trajectory approximations The resulting compressed trajectories can be particularly useful in efficient online or offline post-processing (eg, mobility analytics, statistics, pattern mining, etc) while also facilitating their comparison irrespectively of differing update frequencies

Offline Trajectory Analytics

Abstract: In recent years, there has been observed an “explosion” of trajectory data production due to the proliferation of GPS-enabled devices, such as mobile phones and tablets This massive-scale data generation has posed new challenges in the data management community in terms of storing, querying, analyzing, and extracting knowledge out of such data Knowledge discovery out of mobility data is essentially the goal of every mobility data analytics task Especially in the maritime and aviation domains, this relates to challenging use-case scenarios, such as discovering valuable behavioral patterns of moving objects, identifying different types of activities in a region of interest, environmental fingerprint, etc In order to be able to support such scenarios, an analyst should be able to apply, at massive scale, several knowledge discovery techniques, such as trajectory clustering, hotspot analysis, and frequent route/network discovery methods