Panagiotis Pantazopoulos

Efficient social-aware content placement in opportunistic networks

As content provisioning becomes the driving application of todays (opportunistic) networking environments and the User Generated Content explodes, the problem of devising scalable approaches to placing it optimally within a networking structure becomes more important and challenging. Since the well-known k-median optimization problem that is typically formulated to address it requires global topology and demand information, different approaches are sought for. The latter is the focus of this paper that aims at exploiting social structures, present in emerging networking environments, in order to devise a scalable approach to the optimal or near-optimal content placement. A new metric that captures the nodes social significance or potential for helping establish paths between nodes is introduced and serves as the basis for creating a small scale network sub-graph over which the small-scale content placement problem is solved sequentially until the optimal or near-optimal location is identified. The trade-off between the sub-graphs size and the degree of convergence to the optimal solution is studied through simulations on E-R and B-A random graphs and the effectiveness of the proposed approach is demonstrated.

Comparative assessment of centrality indices and implications on the vulnerability of ISP networks

The position of the nodes within a network topology largely determines the level of their involvement in various networking functions. Yet numerous node centrality indices, proposed to quantify how central individual nodes are in this respect, yield very different views of their relative significance. Our first contribution is then an exhaustive survey and categorization of centrality indices along several attributes including the type of information (local vs. global) and processing complexity required for their computation. We next study the seven most popular of those indices in the context of Internet vulnerability to address issues that remain under-explored in literature so far. First, we carry out a correlation study to assess the consistency of the node rankings those indices generate over ISP router-level topologies. For each pair of indices, we compute the full ranking correlation, which is the standard choice in literature, and the percentage overlap between the k top nodes. Then, we let these rankings guide the removal of highly central nodes and assess the impact on both the connectivity properties and traffic-carrying capacity of the network. Our results confirm that the top-k overlap predicts the comparative impact of indices on the network vulnerability better than the full-ranking correlation. Importantly, the locally computed degree centrality index approximates closely the global indices with the most dramatic impact on the traffic-carrying capacity; whereas, its approximative power in terms of connectivity is more topology-dependent.

How much off-center are centrality metrics for routing in opportunistic networks

The exploitation of social context for routing data in opportunistic networks is a relatively recent trend. Node centrality metrics, such as the betweenness centrality, quantify the relaying utility of network nodes and inform routing decisions, resulting in better performance than more naive routing approaches. Nevertheless,centrality-based routing is far from optimal for three main reasons: a) routing decisions are greedy and message destination-agnostic; b) its performance is highly sensitive to the contact graph over which the node centrality values are computed; c) the global network centrality values have for practical reasons to be approximated by their egocentric counterparts. Our paper experimentally assesses the impact of these three factors on the efficacy of centrality-based routing. Five centrality-based routing variants are compared with each other and against two schemes representing extreme instances of DTN routing complexity: the simple probabilistic forwarding protocol and an ideal scheme with perfect knowledge of future contacts that computes optimal message space-time paths over a novel graph construct with contacts as vertices and time-weighted edges. The results of this comparison are not always inline with intuition and indicate inherent weaknesses of centrality-based routing.

Challenges for Automated Cooperative Driving: The AutoNet2030 Approach

Automated driving is expected to significantly contribute to future safe and efficient mobility. Whereas classical automated approaches solely consider the host vehicle, AutoNet2030 aims to investigate a cooperative approach where communication is used to build decentralized control systems, facilitate cooperative traffic flow optimization, and enhance perception. This chapter introduces the concepts and methodology of AutoNet2030 in order to contribute to a cost-optimized and widely deployable automated driving technology.

Low-cost enhancement of the Intra-domain Internet robustness against intelligent node attacks

Internet vulnerability studies typically consider highly central nodes as favorable targets of intelligent (malicious) attacks. Heuristics that use redundancy adding k extra links in the topology are a common class of countermeasures seeking to enhance Internet robustness. To identify the nodes to be linked most previous works propose very simple centrality criteria that lack a clear rationale and only occasionally address Intra-domain topologies. More importantly, the implementation cost induced by adding lengthy links between nodes of remote network locations is rarely taken into account. In this paper, we explore cost-effective link additions in the locality of the targets having the k extra links added only between their first neighbors. We introduce an innovative link utility metric that identifies which pair of a targets neighbors aggregates the most shortest paths coming from the rest of the nodes and therefore could enhance the network connectivity, if linked. This metric drives the proposed heuristic that solves the problem of assigning the link budget k to the neighbors of the targets. By employing a rich Intra-domain networks dataset we first conduct a proof-of-concept study to validate the effectiveness of the metric. Then we compare our approach with the so-far most effective heuristic that does not bound the length of the added links. Our results suggest that the proposed enhancement can closely approximate the connectivity levels the so-far winner yields, yet with up to eight times lower implementation cost.