[Paper Review] A Novel Framework for Online Amnesic Trajectory Compression in Resource-constrained Environments

Jiajun Liu, Kun Zhao+5 more|arXiv (Cornell University)|May 8, 2016

Data Management and Algorithms32 references3 citations

TL;DR

This paper proposes the Amnesic Bounded Quadrant System (ABQS), an online, constant-complexity framework for error-bounded trajectory compression in resource-constrained environments. By leveraging a convex-hull-based Bounded Quadrant System (BQS) with progressive and fast variants, ABQS enables real-time compression with guaranteed error bounds while intelligently aging data to prevent hard data loss, achieving up to 400 times lower error than baselines under extreme storage constraints.

ABSTRACT

State-of-the-art trajectory compression methods usually involve high space-time complexity or yield unsatisfactory compression rates, leading to rapid exhaustion of memory, computation, storage and energy resources. Their ability is commonly limited when operating in a resource-constrained environment especially when the data volume (even when compressed) far exceeds the storage limit. Hence we propose a novel online framework for error-bounded trajectory compression and ageing called the Amnesic Bounded Quadrant System (ABQS), whose core is the Bounded Quadrant System (BQS) algorithm family that includes a normal version (BQS), Fast version (FBQS), and a Progressive version (PBQS). ABQS intelligently manages a given storage and compresses the trajectories with different error tolerances subject to their ages. In the experiments, we conduct comprehensive evaluations for the BQS algorithm family and the ABQS framework. Using empirical GPS traces from flying foxes and cars, and synthetic data from simulation, we demonstrate the effectiveness of the standalone BQS algorithms in significantly reducing the time and space complexity of trajectory compression, while greatly improving the compression rates of the state-of-the-art algorithms (up to 45%). We also show that the operational time of the target resource-constrained hardware platform can be prolonged by up to 41%. We then verify that with ABQS, given data volumes that are far greater than storage space, ABQS is able to achieve 15 to 400 times smaller errors than the baselines. We also show that the algorithm is robust to extreme trajectory shapes.

Motivation & Objective

• Address the challenge of real-time trajectory compression in memory-, energy-, and storage-constrained devices, such as wildlife tracking sensors with only 4 KB RAM and 1 MB storage.
• Overcome limitations of existing online compression algorithms that suffer from high time/space complexity (e.g., O(n log n) or O(n²)) or require full data buffering.
• Design a framework that supports graceful degradation over time by intelligently managing data aging without hard data loss, even when total data volume exceeds storage capacity.
• Enable high compression efficiency and low error rates under unknown or long operational durations, where data size cannot be estimated in advance.
• Develop a scalable, online solution that maintains constant O(1) space and O(n) time complexity per data point, suitable for embedded and low-power platforms.

Proposed method

• Propose the Bounded Quadrant System (BQS), a novel online compression algorithm that uses a convex hull formed by a bounding box and angular constraints to enable fast, error-guaranteed compression decisions without iterative error calculation.
• Introduce three variants: standard BQS, Fast BQS (FBQS) for O(1) space and O(n) time complexity, and Progressive BQS (PBQS) that incorporates uncertainty vectors for adaptive compression over time.
• Design the Amnesic Bounded Quadrant System (ABQS) as a framework that manages a fixed storage budget by applying an amnesic function that reduces the importance of older data points over time.
• Use a significance function to dynamically adjust error tolerance based on data age, allowing ABQS to prioritize recent data and avoid hard overwriting while maintaining high fidelity for recent trajectories.
• Implement a bounded error mechanism where compression decisions are made using geometric bounds derived from the convex hull, eliminating the need for full buffer scans or maximum deviation calculations.
• Integrate the PBQS component into ABQS to allow progressive refinement of compression, enabling the system to adaptively trade off precision and storage usage over time.

Experimental results

Research questions

• RQ1 Can a trajectory compression algorithm achieve constant space and time complexity (O(1) space, O(n) time) while maintaining error bounds in real-time streaming scenarios?
• RQ2 How can a system manage limited storage over long operational durations without hard data loss, especially when the total data volume exceeds available space?
• RQ3 To what extent does the ABQS framework outperform existing online compression methods in terms of compression error and operational longevity under resource constraints?
• RQ4 How robust is the BQS family to extreme trajectory shapes and varying error tolerance settings in real-world and synthetic data?
• RQ5 Can the progressive compression capability of PBQS enable better long-term performance than fixed-tolerance algorithms when data volume is unknown?

Key findings

• ABQS achieves up to 45% higher compression rates than state-of-the-art algorithms, significantly reducing storage and energy consumption in resource-constrained environments.
• The operational time of target hardware platforms is extended by up to 41% due to reduced memory and computational load from efficient compression.
• Under data volumes far exceeding storage capacity, ABQS achieves 15 to 400 times smaller synchronized errors than baseline methods like FBQS and DP.
• ABQS maintains consistent performance across varying data sizes, with synchronized error around 1.5 km and deviation of 200 m, regardless of input size from 10,000 to 80,000 points.
• The framework is robust to extreme trajectory shapes and error tolerance settings, with minimal performance degradation even when tolerance varies from 10m to 80m.
• ABQS outperforms FBQS and DP significantly after a cutoff point (~27,000 points), where the latter two suffer dramatic error spikes (from 80m to 100km) due to hard data loss, while ABQS maintains stable, low error throughout.