National and Kapodistrian University of Athens
Προσαρμοστική επιδημική διάδοση σε ασύρματα αδόμητα δίκτυα
Abstract
dc:descriptionThe focus of this research is epidemic information dissemination in wireless ad hocnetworks in an adaptive manner. The main targets are the compromise betweendemands for broad information dissemination and reduced energy cost, the assessmentof an optimal solution to the problem and also the achievement of optimal performancein the aforementioned areas with high quality inforamation.Adaptive epidemic dissemination (AED) in ad hoc networks in a noisy environment isa problem with many parameters.Using cross-layering based on channel state-awareness, one can engineer adaptivedissemination sche- mes, which save energy while retaining a fast and effective infection([1] ).This type of scheme [1] is reactive, but a flavor of proactive scheme can be devisedthat uses utility functions in order to adapt transmission characteristics. The proposedsolution is a proof of concept of this idea. With this approach energy cost is savedwhile infection remains successful [2].The above scheme can be enhanced through the use of optimization. Assessingwhich adaptation is most beneficial at every round as in [2], a ”myopic” visibility tothe future is at least achieved. Using optimal stopping theory (OST) one may obtainlonger visibility and approach optimal solutions. The actual optimality depends onsatisfaction of optimal stopping conditions. If these are not fully satisfied, then theobtained solutions are not necessarily optimal. This is displayed in [3] through theuse of the one-stage-look-ahead, (1sla) optimal stopping rule. Improvement over non-adaptive schemes in terms of energy cost save is possible. Broad network infection isstill as effective [3].Optimization is investigated in the periodic broadcast problem, which is addressed witha transition to a near-periodic framework where scheduling is performed using the sametoolcase, that is cross-layer utility functions and an optimal stopping mechanism. Thisproblem is modeled as a secretary problem where the use of OST offers optimality.Compared against non-OST [2] and non-adaptive schemes this scheme wins in theenergy cost save area and allows for successful network infection. The energy costof channel state information (CSI) retrieval may also be considered. In this case, thesystem converges to a state where dissemination cost is dramatically reduced whilea large proportion of the residual, steady-state energy cost is due to CSI acquisition.Exact relations depend on parameters of the problem [4].Furthermore, the possibility to improve information quality is investigated using schemessuch as the one in [4]. Information freshness is originally considered as the measureof quality. Adaptation of transmission characteristics based on information freshnessis suggested aiming to improve the average age of the infecting information. In thismanner, the idea of adapting transmission characteristics based on the actual payloadof the infecting information is introduced.Tuning broadcasts down to a polite gossip during the route discovery phase in routingprotocols is an additional attractive area. This problem is addressed within the frameworkof the AODV protocol. The tradeoff between energy cost save and fast routing informationdiscovery emerges here. In an effort to expand the technique to the actual routed datatransmission (i.e. the user data), we move from broadcasting to a unicast landscape.Here, the tuning down of transmissions may have significant impact on informationpropagation. It is shown that the problem is best addressed using metrics reflecting aconceived quality of service and by extension quality of information.The time scheduling problem enhanced by the additional information quality requirementcan be formally expressed as an optimal stopping problem with known finite horizon.The quality of the information may be defined based on its properties (such as freshnessas mentioned earlier) or parameters of the protocol concerned with the dissemination,such as TTL in case of AODV.In this manner, information quality emerges as a third component besides energy costand network infection, which participates the addressed tradeoff. Depending on thedefined information quality seeking a compromise between these three requirementsis possible.
Degree
thesis:*- Grantor dc:publisher
- National and Kapodistrian University of Athens
- Year dc:date
- 2018
Author and committee
dc:creator, dc:contributor.*- Authors dc:creator
-
- Kontos, Theofanis
- Κοντός, Θεοφάνης
Subjects
dc:subject × 12Rights
- Language dc:language
- gre
Identifiers
dc:identifier.*- Identifier
- 10.12681/eadd/45651