Back to search

University of Thessaly (UTH)

Parallel query processing on big spatial data

Abstract

dc:description

Nowadays we are storing overwhelming volumes of Spatial data. The continuously growing need for processing such big data has led us create distributed or parallel frameworks, to address the emerging computational overhead. Such data (often termed as geospatial data), are used in many geographically enabled applications. To exploit these data, efficient processing of spatial queries is of great importance due to the wide area of applications that may address such queries. The most common spatial queries where points are involved are point-location, window, distance-range and k nearest-neighbor queries (k-NNQs). At a higher level, such queries have been used as the basis of many complex operations in advanced applications, for example, geographical information systems (GIS), location-based systems (LBS), geometric databases, CAD, etc.The spatial queries processing performance is usually affected by two major factors. The computational and the storage factor. The first one can be addressed by many architectures,for example single core (sequential execution), multi-core (parallel execution), multi-node or even GPU processing models. The second factor is also of great importance and affects the retrieval or storage of information. The advent of non-volatile memories (NVM) has enabled a brand-new class of storage devices with exciting features that will prevail in the storage market in the near future. Their high read and write speeds, small size, low power consumption and shock resistance are some of the reasons that made them storage medium of choice.NAND flash is undoubtedly the most popular NVM today. Storage devices based on NAND Flash are found both in consumer devices and enterprise data-centers.The increasing needs for efficient storage drove the emergence of Solid-State Drives (SSDs). Aside from the design and development of the above spatial queries, we furthermore tested their efficiency in Edge Computers. Edge computing is a distributed computing architecture where computation and data storage is as close to the source of data as possible. We also designed an Edge computing and IoT distributed architecture, where we described thoroughly the way we can exploit the dynamics of our techniques.In this thesis we focus on the design and implementation of spatial queries that take advantage of multi-core CPU, GPU and SSD for storage. We employ techniques for faster and fewer computations, computation reductions as well as data processing reductions, using spatial locality and data distribution, on either synthetic or real data. Based on these techniques we develop efficient spatial query algorithms using multi-core CPU, we develop efficient spatial query algorithms using GPU, we exploit the massive I/O advantages of SSDs, we apply these queries to Edge Computers, we perform extensive experimental tests to compare our algorithms to other existing ones using synthetic and real spatial data.

Degree

thesis:*
Grantor dc:publisher
University of Thessaly (UTH)
Year dc:date
2022

Author and committee

dc:creator, dc:contributor.*
Authors dc:creator
  • Velentzas, Polychronis
  • Βελέντζας, Πολυχρόνης

Subjects

dc:subject × 22

Rights

Language dc:language
eng

Identifiers

dc:identifier.*
Identifier
10.12681/eadd/52863

Chain of custody

source
Harvested from
Greek National Archive of PhD Theses
Base URL
phdtheses.ekt.gr/eadd_oai/request
Last updated
2026-11-18
Source record
OAI-PMH GetRecord
citation

Velentzas, Polychronis; Βελέντζας, Πολυχρόνης. Parallel query processing on big spatial data. University of Thessaly (UTH), 2022. http://hdl.handle.net/10442/hedi/52863