SAVING as Data Storage: A Social Choice Game for Urban Cache Recruitment 

Get Complete Project Material File(s) Now! »

Socially Aware Vehicular Information-centric NetworkinG (SAVING)

SAVING is a data storage framework to find and recruit vehicles for collaborative urban caching. This is realized thanks to the following interlinked contributions:
• A content spatio-temporal availability-popularity relation is proposed to decide a content importance in order to be cached at a vehicle.
• A vehicle eligibility metric “GRank” is presented for social aware content caching allowing a vehicle to classify its caching capability proportionally to its connectivity, cost and reachability in the network.
• An incentive driven social welfare game is formulated to fairly select among the best ranked vehicles the eligible candidates to cache content for different urban neighborhoods catering individual rationality.
• An optimization problem is modeled following by a vehicle selection algorithm to ensure the selected vehicles satisfy urban content availability requirements for a given budget.
• In order to even further optimize the caching, a coalition game is proposed to form mobile fogs [50] for resource pooling at spatio-temporally co-located vehicles where service provider selects coalitions as urban caches to maximize content availability.

Thesis Organization

The remaining of the document is organized as follows. The following chapter highlights the background along a review on the state of the art data collection and storage schemes in the literature. The proposed data collection approach VISIT is presented by first defining novel metrics to classify vehicle eligibility in Chapter 3 followed by an optimized selection of the best set of vehicles to collect data from urban streets in Chapter 4. Chapters 5 and 6 deals with the data storage scheme SAVING where the identification and recruitment of individual urban information hubs is addressed in Chapter 5 while the collaborative distributed caching among vehicles is described in Chapter 6. The Chapter 7.1 concludes our work with a discussion on future insight in Chapter 7.2.

ICN meets Urban Data Collection

Urban sensing and vicinity monitoring using vehicles has attracted lots of researchers in the past few years and several schemes are proposed [27] [44], where sensor-equipped vehicles sense and share data in a vehicular network. For example, authors in [8] focus on the collection of multimedia data from urban streets using vehicles present on roads. Another example is CarTel [17] which is a distributed sensor communication system designed to gather, visualize and send data form vehicle-embedded sensors.
CarSpeak [25] is another example which allows vehicle to collaborate and access sensory information captured by neighboring vehicles in the same manner as it can access its own content. All these preliminary approaches proposed architectures and general frameworks to adapt ICN to vehicular network. Optimizing the data collection and storage was not specifically addressed.
Recruitment of such vehicles for urban sensing is being studied only since recently. For example, in [13], participants with high reputation are recruited to perform urban sensing. The idea is to cover an area of interest with a limited budget, however, the coverage metric is confined to particular road sections with the limitations of utilizing the infrastructure network. Moreover, the authors do not provide any metric to classify and identify the important participants. Another approach suggests that the ICN paradigm is as a promising solution to cater the peculiarities of the vehicular environment, characterized by dynamic topologies, unreliable broadcast channels, short-lived and intermittent connectivity [40]. Similarly, the authors in [54] studied the coverage problem for urban vehicular sensing where a metric called as Inter-Cover time is proposed to characterize the coverage opportunities as well as assess the coverage quality. Based on that, a vehicle selection algorithm is proposed to select the minimum number of vehicles to achieve the required coverage quality requirements. In both cases, the authors only considered coverage quality as the metric while ignoring the vehicle inherent abilities such as its availability and connectivity to facilitate data collection in an urban scenario. The above coverage metrics are unable to consider spatio-temporal coverage for the vehicles availability for the data collection and storage at different locations and times.

Social Network meets Vehicular Network

Identification of influential information hubs for publishing/spreading information is required in applications such as social networks. Another interesting application is found in medical sciences to find epidemic disease spreaders [24]. Similarly, Google’s PageRank [33] algorithm ranks the importance of a web-page in an Internet search based on the number of web links directed towards it. More generally, Social Network Analysis (SNA) [36] is required to identify important nodes in a social network usually relying on well known network centrality schemes such as Degree, Closeness, Betweenness and Eigenvector centrality.
Degree centrality considers the number of direct (one hop) neighbors of a node, where Closeness centrality is the inverse of the sum of the lengths of the shortest paths from a node to the rest of the nodes in the network. Betweenness centrality is the fraction of all pairs of shortest paths passing through a node, where Eigenvector centrality is the node’s influence measure in the network [7]. By tweaking these centrality measures, algorithms such as BubbleRap [16] and ML-SOR[38] are proposed, where nodes with high centrality score are preferred for data dissemination and routing in Opportunistic Social Networks. Another important work [4] suggests VIP delegation to offload network traffic based on opportunistic contacts. The authors consider well known social network attributes (betweenness, closeness, degree centrality and pagerank) to select delegate nodes in an urban area according to two methods; global (network-based) and hood (community-based) selection.

READ  Experiences of the boundary between work and home life

Table of contents :

Acknowledgments
Abstract
Publications
1 Introduction 
1.1 Motivation
1.2 Problem Statement
1.3 Contribution
1.3.1 Vehicular Information-centric Socially Inspired Telematics (VISIT)
1.3.2 Socially Aware Vehicular Information-centric NetworkinG (SAVING)
.1.4 Thesis Organization
2 State of the Art 
2.1 Background: Information Centric Networking
2.2 Data Collection using Vehicles
2.2.1 ICN meets Urban Data Collection
2.2.2 Social Network meets Vehicular Network
2.2.3 Discussion
2.3 Data Storage in a Vehicular Network
2.3.1 Recruitment of Vehicles as Content Caches
2.3.2 Content Cache Management
2.3.3 Social Networks meet Caching
2.3.4 Discussion
2.4 Conclusion
3 VISIT for Data Collection: Novel Centrality Metrics to Identify Eligible Candidates 
3.1 Introduction
3.2 Context and Motivation
3.3 InfoRank: An Information Importance Based Centrality Scheme
3.3.1 Network Model
3.3.2 User Interests Satisfaction
3.3.3 Information Validity Scope
3.3.4 InfoRank Computation
3.4 CarRank: A Vehicle Centrality Algorithm
3.4.1 Information Importance
3.4.2 Spatio-temporal Availability
3.4.3 Neighborhood Importance
3.4.4 CarRank Computation
3.5 Performance Evaluation
3.5.1 Simulation Scenario
3.5.2 Results: Individual Vehicle Ranking
3.5.2.1 Cumulative Satisfied Interests
3.5.2.2 Temporal behavior analysis of the top nodes
3.5.2.3 Throughput
3.5.2.4 ICN Evaluation – In-Network Caching
3.5.3 Discussion
3.6 Conclusions
4 VISIT for Data Collection: An Optimal Vehicles Selection Scheme 
4.1 Introduction
4.2 Context and Motivation
4.3 Recruitment of Optimal Vehicles for Efficient Road Sensing (ROVERS)
4.3.1 Problem Formulation
4.3.1.1 Spatio-temporal Coverage
4.3.1.2 Vehicle Centrality
4.3.1.3 Dedicated Budget
4.3.2 Algorithm: Optimized Set of Vehicles selection
4.4 Performance Evaluation
4.4.1 Results: Best set of vehicles selection
4.4.1.1 Cumulative Satisfied Interests
4.4.1.2 Throughput
4.5 Conclusions
5 SAVING as Data Storage: A Social Choice Game for Urban Cache Recruitment 
5.1 Introduction
5.2 Context and Motivation
5.3 Autonomous Information Hub Identification – GRank
5.3.1 Information Global Centrality
5.3.2 GRank Computation
5.3.3 Summary
5.4 Information Hubs Recruitment
5.4.1 Game Formulation
5.4.2 Vehicle Utility as a Social Norm
5.4.3 Social Welfare Optimization
5.4.4 Algorithm: Vehicle Selection as Content Caches
5.5 Performance Evaluation
5.5.1 Simulation Scenario
5.5.2 Results: Individual Vehicles Ranking
5.5.2.1 Cumulative Satisfied Interests
5.5.2.2 Temporal Network Behavior Analysis
5.5.2.3 Aggregated Per Node Throughput
5.5.2.4 In-network Cache Hit-rate
5.5.3 Results: Selected Set of Vehicles
5.5.3.1 Success Rate
5.5.3.2 Aggregated Throughputs
5.5.3.3 In-network Cache Hit-Rate
5.6 Conclusions
6 SAVING as Data Storage: A Collaborative Caching Game at Mobile Fogs 
6.1 Introduction
6.2 Context and Motivation
6.3 Who should cache What?
6.3.1 Network Model
6.3.2 Spatio-temporal Content Profile
6.3.3 Node Eligibility
6.3.3.1 Local Centrality
6.3.3.2 Global Centrality
6.4 Distributed Fogs Formation as Coalition Game
6.4.1 Solution: The Core
6.4.2 Algorithm: Optimal Selection among Coalitions
6.5 Performance Evaluation
6.5.1 Simulation Scenario
6.5.2 Simulation Results
6.5.2.1 Cache Hits
6.5.2.2 Offload Benefit
6.5.2.3 Spatio-temporal coalitions
6.5.3 Summary of Findings
6.6 Conclusions
7 Conclusions and Future Work 
7.1 Conclusions
7.2 Future Work
List of Figures
List of Tables
Bibliography

GET THE COMPLETE PROJECT

Related Posts