Space and time are key elements for many computer-based systems and often elevated to first-class abstractions. In tuple-based coordination, Linda primitives have been independently extended with space (with tuples and queries spanning spatial regions) or time information (mostly for tuple scoping). However, recent works in collective adaptive systems and aggregate computing show that space and time can naturally be considered as two intertwined facets of a common coordination abstraction for situated distributed systems. Accordingly, we introduce the Spatiotemporal Tuples model, a natural adaptation of Linda model for physically deployed large-scale networks. Unlike prior research, spatiotemporal properties – expressing where and when a tuple should range and has to be deposited/retrieved – naturally turn into specifications of collective adaptive processes, to be carried on in cooperation by the devices filling the computational environment, and sustaining tuple operations in a resilient way, possibly even in mobile and faulty environments. Additionally, the model promotes decentralised implementations where tuples actually reside where they are issued, which is good for supporting peer-to-peer and mobile ad-hoc networks as well as privacy. In this paper, we (i) present and formalise the Spatiotemporal Tuples model, based on the unifying notion of computational space-time structure, (ii) provide an implementation in the ScaFi aggregate computing framework, turning tuple operations into aggregate processes, and finally (iii) provide evaluation through simulation and a rescue case study.

Tuple-Based Coordination in Large-Scale Situated Systems

Audrito G.
2021-01-01

Abstract

Space and time are key elements for many computer-based systems and often elevated to first-class abstractions. In tuple-based coordination, Linda primitives have been independently extended with space (with tuples and queries spanning spatial regions) or time information (mostly for tuple scoping). However, recent works in collective adaptive systems and aggregate computing show that space and time can naturally be considered as two intertwined facets of a common coordination abstraction for situated distributed systems. Accordingly, we introduce the Spatiotemporal Tuples model, a natural adaptation of Linda model for physically deployed large-scale networks. Unlike prior research, spatiotemporal properties – expressing where and when a tuple should range and has to be deposited/retrieved – naturally turn into specifications of collective adaptive processes, to be carried on in cooperation by the devices filling the computational environment, and sustaining tuple operations in a resilient way, possibly even in mobile and faulty environments. Additionally, the model promotes decentralised implementations where tuples actually reside where they are issued, which is good for supporting peer-to-peer and mobile ad-hoc networks as well as privacy. In this paper, we (i) present and formalise the Spatiotemporal Tuples model, based on the unifying notion of computational space-time structure, (ii) provide an implementation in the ScaFi aggregate computing framework, turning tuple operations into aggregate processes, and finally (iii) provide evaluation through simulation and a rescue case study.
2021
23rd IFIP WG 6.1 International Conference on Coordination Models and Languages, COORDINATION 2021 held as part of 16th International Federated Conference on Distributed Computing Techniques, DisCoTec 2021
Online
2021
Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Springer Science and Business Media Deutschland GmbH
12717
149
167
978-3-030-78141-5
978-3-030-78142-2
aggregate computing; ScaFi; self-organisation; spatial tuples; tuple-based coordination
Casadei R.; Viroli M.; Ricci A.; Audrito G.
File in questo prodotto:
File Dimensione Formato  
paper-2021-coordination-tuples.pdf

Open Access dal 02/10/2023

Tipo di file: POSTPRINT (VERSIONE FINALE DELL’AUTORE)
Dimensione 561 kB
Formato Adobe PDF
561 kB Adobe PDF Visualizza/Apri

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/2318/1806999
Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus 5
  • ???jsp.display-item.citation.isi??? ND
social impact