The study of mobile entities that based on local information have to accomplish global tasks is of main interest for the scientific community. Classic models for the activation and synchronization of mobile entities are the fully-synchronous (FSYNC), semi-synchronous (SSYNC), and asynchronous (ASYNC) models, where entities alternate between active and inactive states with different timing. According to the assumed synchronization model, very different results have been achieved in the field of distributed computing. One of the main outcomes is the big gap between the ASYNC and the other models in terms of manageability and algorithm design. In fact, there are still many problems for which it is not known whether synchronicity is crucial for designing resolution algorithms or not. In order to better understand the ASYNC case, here we propose a further model referred to as the semi-asynchronous (SASYNC). This slightly deviates from SSYNC. In fact, like in SSYNC (and FSYNC), the duration of the activation of an entity is kept of fixed time whereas, like in ASYNC, the starting instant of the activation is not fully synchronized with the possible activation of other entities. We show that for entities moving on graphs, the SSYNC model allows accomplishing more tasks than the SASYNC that in turn allows accomplishing more tasks than the ASYNC. Furthermore, our results show that, especially to tackle problems in the Euclidean plane, the SASYNC model is already quite challenging, therefore there is no need to get involved with complications arising in the ASYNC model.

“Semi-Asynchronous”: a new scheduler in distributed computing

Cicerone S.;Di Stefano G.;Navarra A.
2021-01-01

Abstract

The study of mobile entities that based on local information have to accomplish global tasks is of main interest for the scientific community. Classic models for the activation and synchronization of mobile entities are the fully-synchronous (FSYNC), semi-synchronous (SSYNC), and asynchronous (ASYNC) models, where entities alternate between active and inactive states with different timing. According to the assumed synchronization model, very different results have been achieved in the field of distributed computing. One of the main outcomes is the big gap between the ASYNC and the other models in terms of manageability and algorithm design. In fact, there are still many problems for which it is not known whether synchronicity is crucial for designing resolution algorithms or not. In order to better understand the ASYNC case, here we propose a further model referred to as the semi-asynchronous (SASYNC). This slightly deviates from SSYNC. In fact, like in SSYNC (and FSYNC), the duration of the activation of an entity is kept of fixed time whereas, like in ASYNC, the starting instant of the activation is not fully synchronized with the possible activation of other entities. We show that for entities moving on graphs, the SSYNC model allows accomplishing more tasks than the SASYNC that in turn allows accomplishing more tasks than the ASYNC. Furthermore, our results show that, especially to tackle problems in the Euclidean plane, the SASYNC model is already quite challenging, therefore there is no need to get involved with complications arising in the ASYNC model.
File in questo prodotto:
File Dimensione Formato  
09373406.pdf

accesso aperto

Descrizione: Articolo principale
Tipologia: Documento in Post-print
Licenza: Creative commons
Dimensione 1.05 MB
Formato Adobe PDF
1.05 MB Adobe PDF Visualizza/Apri
Pubblicazioni consigliate

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/11697/162311
Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus 11
  • ???jsp.display-item.citation.isi??? 7
social impact