University of Sussex
Browse
Al-Eidarous, Walla.pdf (5.23 MB)

Efficient opportunistic routing in dense mobile networks

Download (5.23 MB)
thesis
posted on 2023-06-09, 19:09 authored by Walla Al-Eidarous
The usage of smartphones is nowadays ubiquitous. Their simultaneous support for longand short-range communication has enabled the deployment of opportunistic, device-todevice networks, which exploit human mobility to enable and facilitate communication and content exchange among peer devices. Devices connect to each other without human intervention, potentially with the assistance of the cellular network provider. The underlying network topology constantly changes, depending on the mobility patterns of the participating mobile devices. Mobile devices support various technologies for discovering their location; GPS is very accurate but it works only outdoors and is power-hungry, whereas location discovery based on nearby announced SSIDs and/or the current cell ID is less accurate but power-friendly. Indoor localisation is much more challenging; approaches that are based on inertial sensors and dead reckoning, along with deployed beacons and pre-calculated signal strength maps have been proposed. In this thesis, we develop GeoHawk, a routing protocol for dense mobile networks that support opportunistic communication and content dissemination among mobile devices in crowded events. The driving use case has been the Grand Mosque, the largest mosque in the world located at the heart of the city of Makkah in Saudi Arabia. During the Ramadan and Hajj, viii the Grand Mosque can get extremely crowded, with anticipated number of visitors close to 2.5 million, after the current expansion work is completed. The proposed protocol incorporates a novel distributed localisation technique that can be used in conjunction with the protocol, when GPS is not available. GeoHawk deals with the very high density of users/devices by heavily aggregating routing information using Bloom filters. Identifiers of mobile devices that reside within specific geographical regions are disseminated in the network in the form of Bloom filters. Said geographical regions are dynamically created and destroyed; their size evolves to reflect the uncertainty in the topology, due to mobility and potential inaccuracies of the underlying location estimation mechanism. Bloom filters are also decayed to reflect information ageing. Devices exchange routing information with their neighbours and announce aggregated information (i.e. Bloom filters) in messages that propagate towards specific directions and reach distant areas of the opportunistic network. Data is then disseminated (and replicated through a simple but efficient ticketing mechanism) towards directions where the information about the existence of the destination node is stronger. Upon reaching the best-known region for the destination node, a message is either flooded, if the belief that the node resides in the region is strong (as indicated by a belief threshold), or, in the opposite case, redirected to a randomly selected region. The distributed localisation algorithm is a novel synthesis of existing techniques, including Pedestrian Dead Reckoning, estimated location sharing and particle filtering. Our approach can provide reasonable errors in the estimation, which allow the routing protocol to effectively deliver messages to destination nodes. We evaluate GeoHawk using extensive experimentation in the ONE simulator. We have developed mobility models that approximate the user behaviour in the targeted use ix cases and communication environments. We have experimented with a large variety of configuration parameters that affect the behaviour of the proposed protocol and recorded its performance in terms of message delivery ratio and latency as well as induced network overhead. We show that the GeoHawk’s performance is superior to baseline protocols, namely Epidemic, PRoPHET and WSR.

History

File Version

  • Published version

Pages

193.0

Department affiliated with

  • Informatics Theses

Qualification level

  • doctoral

Qualification name

  • phd

Language

  • eng

Institution

University of Sussex

Full text available

  • Yes

Legacy Posted Date

2019-09-25

Usage metrics

    University of Sussex (Theses)

    Categories

    No categories selected

    Exports

    RefWorks
    BibTeX
    Ref. manager
    Endnote
    DataCite
    NLM
    DC