Self-organisation in traffic Signal control algorithms

Movius, Samantha Jane (2018-03)

Thesis (PhD)--Stellenbosch University, 2018.

Thesis

ENGLISH ABSTRACT: Two popular types of traffic signal control are fixed-time control and vehicle-actuated control. The latter method involves switching traffic signals based on detected traffic ows and thus offers more exibility than the former, which relies solely on cyclic, predetermined signal phases. The notion of self-organisation has relatively recently been proposed as an alternative approach towards improving traffic signal control, particularly during periods of light traffic flow, due to its flexible nature and its potential to result in emergent behaviour. The effectiveness of five self-organising traffic signal control strategies from the literature, as well as a fixed-time control strategy, have previously been compared in a simulated environment. Various shortcomings of three of these algorithms are pointed out in this dissertation and algorithmic improvements are suggested to remedy these defficiencies. The significant improvements resulting from these algorithmic modifications are then quantified by means of their implementation in a newly designed agent-based, microscopic traffic simulation model. Two novel self-organising traffic control algorithms are also proposed in this dissertation. These algorithms have been designed in such a way as to avoid certain shortcomings discovered in the aforementioned algorithms. The two novel algorithms, together with the improved versions of the three existing algorithms and the remaining pair of algorithms from the literature, are also subjected to thorough testing in the aforementioned simulation framework in terms of their propensity to facilitate the formation of green waves and to recover from various disruptions (such as road closures or abnormal traffic induced by large events) within the context of both gridded street networks and corridors with approaching side roads. All eight algorithms are finally implemented in a simulation model representing an existing road network in order to compare and evaluate the effectiveness of these algorithms within the context of a real-world scenario. It is found that the two newly proposed algorithms outperform existing self-organising traffic signal control algorithms under certain traffic conditions and road network topologies.

AFRIKAANSE OPSOMMING: Twee populêre tipes verkeersbeheertegnieke is vaste-tyd beheer en voertuig-geïnduseerde beheer. Volgens laasgenoemde tegniek berus die wisseling van verkeerseine op die meting van verkeersvloei wat meer buigsaamheid bied as die eersgenoemde tegniek wat berus op sikliese, voorafbepaalde seinfases. Die begrip van self-organisering is relatief onlangs as 'n alternatief vir verbeterde verkeerseinbeheer voorgestel as gevolg van die buigsame aard en die potensiaal daarvan om te lei na ontluikende gedrag. Die doeltreffendheid van vyf self-organiserende verkeersbeheerstrategieë uit die literatuur is reeds voorheen in 'n gesimuleerde omgewing vergelyk. Verskeie tekortkominge van drie van hierdie algoritmes word in hierdie proefskrif uitgewys en algoritmiese verbeterings word voorgestel om hierdie tekortkominge uit die weg te ruim. Die beduidende verbeterings wat uit hierdie algoritmiese veranderinge voortspruit, word dan aan die hand van 'n nuut-ontwerpte agent-gebaseerde verkeers-mikrosimulasiemodel gekwantifiseer. Twee nuwe self-organiserende erkeersbeheeralgoritmes word ook in hierdie proefskrif daargestel. Hierdie algoritmes is ontwerp om sommige tekortkominge wat in die bogenoemde algoritmes ontdek is, aan te spreek. Die twee nuwe algoritmes, tesame met die verbeterde weergawes van die drie bestaande algoritmes sowel as die oorblywende paar algoritmes, word ook aan deeglike toetse in die bogenoemde simulasieraamwerk onderwerp in terme van hul vermoë om die vorming van groen golwe te bewerkstellig en om van ontwrigtings (wat deur straatsluitings of abnormale verkeersdruk as gevolg van groot byeenkomste te weeg gebring word) in 'n netwerk van straatblokke en in 'n korridor met aansluitende systrate te herstel. Al agt algoritmes word laastens ook in 'n simulasiemodel van 'n bestaande padnetwerk geïmplementeer om sodoende die doeltreffendheid van die algoritmes in die konteks van 'n realistiese scenario te vergelyk. Daar word in al die simulasie-eksperimente bevind dat die twee nuwe algoritmes die bestaande self-organiserende algoritmes onder ligte verkeerstoestande uitstof, terwyl 'n eenvoudige vaste seinfase-siklus beheerstrategie die beste benadering onder swaar verkeerstoestande is.

Please refer to this item in SUNScholar by using the following persistent URL: http://hdl.handle.net/10019.1/103301
This item appears in the following collections: