Back to search

National Technical University of Athens (NTUA)

Socio-aware content allocation in complex networks via efficient monitoring of information dissemination

Abstract

dc:description

This dissertation focuses on the analysis and design of content placement approaches in complex networked systems, i.e., cyber and cyber-physical networks, via efficient monitoring and inference of its constituents’ interplay. Despite the fact that the types of networks under examination exhibit diverse and distinctive features, the proposed methods share a common objective, namely the utilization of minimal amount of resources in order to track, infer or predict the explicit and implicit interactions among the entities of a network, which are governed by complex constraints. Inspired by methodologies employed to solve problems of coverage in physical networks, such as Wireless Sensor Networks (WSNs), this thesis extends, through Social Network Analysis (SNA), the notion of coverage in cyber and cyber-physical networks. Problems of information tracking and inference, along with influence maximization and content allocation are mapped to covering and packing problems of combinatorial optimization, which are known to be NP-hard. To address the former, heuristic approaches as well as algorithms with provable approximation guarantees are designed and analysed.The concept of coverage is introduced in WSNs as an important metric, which measures how well the network monitors a region of interest. In this thesis, the problem of computing a minimal amount of WSN resources to monitor an obstructed field of interest is formulated as a topology control problem. A framework based on computational geometry is introduced and two algorithms, a centralized and a distributed one, adjust the sensors’ sensing ranges in order to maximize the ratio of covered area to consumed energy, while ensuring a minimum coverage percentage.Monitoring in cyber networks, such as Online Social Networks (OSNs) is the process of tracking the users’ interactions, as captured by the information flow. In networks of billions of users, it is of great significance to reduce the resources required to infer the developing diffusion dynamics. The problem of finding a minimal set of monitoring nodes is mapped to finding a maximal independent set. A greedy approach is developed for its solution, followed by a graph coloring scheme that enables information tracking and a statistical learning technique, which infers the information diffusion graph.Approaches of monitoring or inference of information propagation can be integrated into Recommendation Systems (RSs). These RSs are known as Information Diffusion Aware Recommendation Systems (IDARS). This dissertation treats information diffusion aware recommendations as a content allocation problem with user coverage constraints. An IDARS is designed to find a subset of users to assign them various types of content, such that eventually every user achieves a coverage goal, i.e., all users in the network receive, via information dissemination, at least a minimum number of recommendations, while the total relevance of users to the recommended items is maximized. This is translated to a generalization of the Minimum Weighted Set Cover Problem and a greedy approximation algorithm is proposed for its solution.In Mobile Social Networks (MSNs) and platforms of streaming services the information on users’ features is frequently acquired by RSs. This information along with the users’ mobility patterns are exploited to derive local communities and encourage collaboration in content sharing. According to the users’ physical and social ties and by acknowledging the impact of recommendations in users’ content requests, the problem of content placement at edge caching networks and content sharing via Device-to-Device (D2D) communication is investigated under various objectives. At first, allocating items to a physical network, i.e., a caching network, is mapped to a problem of cache hit ratio maximization and the solution is approximated by a dynamic programming based approach. Next, content allocation is studied in a cyber-physical network formed by a caching network and a platform of streaming services. In particular, D2D-based opportunistic offloading is studied jointly with cache aware recommendations. Multiple criteria based on user mobility patterns are proposed to determine the user equipment participating in the offloading. Expressing the user Quality of Experience (QoE) as a function of user-content relevance and its expected delivery delay, the problem of caching and recommendations is treated as a user QoE maximization problem. It is addressed by a framework that solves sequentially the problems of minimum delay content delivery, content placement in caches and cache-aware recommendations, ensuring that each user will be recommended of highly preferred content with minimum delivery delay.

Degree

thesis:*
Grantor dc:publisher
National Technical University of Athens (NTUA)
Year dc:date
2021

Author and committee

dc:creator, dc:contributor.*
Authors dc:creator
  • Vitoropoulou, Margarita
  • Βιτοροπούλου, Μαργαρίτα

Subjects

dc:subject × 14

Rights

Language dc:language
eng

Identifiers

dc:identifier.*
Identifier
10.12681/eadd/50167

Chain of custody

source
Harvested from
Greek National Archive of PhD Theses
Base URL
phdtheses.ekt.gr/eadd_oai/request
Last updated
2026-11-18
Source record
OAI-PMH GetRecord
citation

Vitoropoulou, Margarita; Βιτοροπούλου, Μαργαρίτα. Socio-aware content allocation in complex networks via efficient monitoring of information dissemination. National Technical University of Athens (NTUA), 2021. http://hdl.handle.net/10442/hedi/50167