dc.contributor.authorAthul Harilal
dc.date.accessioned2016-01-11T02:01:58Z
dc.date.available2016-01-11T02:01:58Z
dc.date.issued2016
dc.identifier.urihttp://hdl.handle.net/10356/65881
dc.description.abstractIn this dissertation, we aim to study the spread of infection in the light of Susceptible Infected Recovered Infected (SIRI) model. In order to achieve the objective, an estimator by the name Heterogeneous Infection Spreading Source (HISS) was developed. The estimator does the task of emulating the spread of infection by defining state space variables and auxiliary variables. Thus the estimator tries to obtain a distribution which is similar to the observed state of nodes. It not only estimates the most likely origin of infection, but also computes the most probable snapshot time. The estimator also incorporates side information. Side information is defined as the prior knowledge of a certain fraction of nodes to be in one of the three states namely Susceptible (S), Infected (I) or Recovered (R). This is observed before the snapshot instance. It is implemented to observe the detection accuracy of the true source with different number of known side information. The simulations are run on random tree graphs of degree 4 and size 1000 and on facebook network of size 500. The performance of our estimator are compared with Dynamic message Passing (DMP) algorithm and Jordan centrality. HISS estimator outperforms both of the other estimators. It accurately identifies the true source over a wide range of infection and reinfection rates.en_US
dc.format.extent90 p.en_US
dc.language.isoenen_US
dc.subjectDRNTU::Engineering::Electrical and electronic engineeringen_US
dc.titleInfection source estimation under the SIRI modelen_US
dc.typeThesis
dc.contributor.supervisorTay Wee Pengen_US
dc.contributor.schoolSchool of Electrical and Electronic Engineeringen_US
dc.description.degreeMaster of Science (Communications Engineering)en_US


Files in this item

Thumbnail

This item appears in the following Collection(s)

Show simple item record