Prométhée Spathis

Virtualizing vehicular node resources

With emerging geo-distributed services, there is a need to coordinate the use of resources offered by field-area networks. In the case of vehicular networks, such resources include the processing, sensing, and storage capabilities offered to service providers for urban sensing or intelligent transportation. In this paper, we propose to virtualize the resources embedded on the vehicular nodes to allow multiple tenants to coexist and deploy their services on the same underlying mobile substrate. Virtualization is the task of an infrastructure provider that controls the mobile substrate and allocates sliced resources to the tenants. A service results from a collection of virtual machines hosted on the mobile nodes allocated by the infrastructure provider. Efficient utilization of the node resources may trigger virtual machine migrations. We study the problem of virtual machine migrations through V2V communications between mobile nodes. To evaluate the impact of such migrations on the resource allocation process, we use the real traces of a bus transit system to simulate a vehicular network where virtual machines migrate via V2V communications. Our results show that virtual machines of several hundreds of Megabytes can migrate between moving buses. We then discuss design principles and research issues toward the full virtualization of opportunistic networks.

Vehicular carriers for big data transfers (Poster)

In the latest years, Internet traffic has increased at a significantly faster pace than its capacity, preventing efficient bulk data transfers such as data-center services and high-definition user-generated content applications. In this paper, we propose to take advantage of the existing worldwide road infrastructure as an offloading channel to help the legacy Internet assuage its burden. Our results suggest that piggybacking data on vehicles can easily lead to network capacity in the petabyte range.

Software-defined vehicular backhaul

The network of roads and highways is a promising candidate to help network operators offload their infrastructure and cope with the ever-growing amount of data exchanged on the Internet. By piggybacking data onto vehicles, roads can be turned into a large-capacity transmission system when considering the increasing number of journeys involving vehicles. The data to be transferred is opportunistically loaded on or off the vehicles at specific locations referred to as offloading spots. Two of the main challenges of such a system are how to assign the road paths matching the data transfer requirements and how much data to allocate to each flow of vehicles. We propose a centralized SDN-like architecture consisting of a central controller acting as a service broker and the offloading spots as SDN agents. The controller computes the road paths that accommodate the data transfer requirements and installs the corresponding forwarding states at each offloading spot along those paths. We describe our SDN-controlled offloading system and evaluate its performance using road traffic counts from France. Our numerical results show that the controller can achieve efficient and fair allocation of multiple data transfers between major cities of France. Each transfer successfully delivers over 10 PB of data within a week when considering that 10% of vehicles on the road are equipped with 1TB of storage.

SERVUS: Reliable low-cost and disconnection-aware broadcasting in VANETs

The future of Vehicular Ad-Hoc Networks (VANET) will rely mainly on the support of efficient information dissemination protocols, whether we talk about safety applications that warn the driver of an imminent collision, a simple update on traffic conditions, or road-side advertisements. One of the greatest challenges when designing such protocols is how to deliver packets efficiently in a highly mobile environments under intermittent connectivity. Surprisingly, this problem has been under-investigated in the literature. In this paper, we propose, design, and evaluate SERVUS1, a robust dissemination protocol that guarantees packet propagation with high delivery ratio and low overhead. SERVUS includes a new broadcast management mechanism that takes advantage of the inherent behavioral properties of the VANET environment. In particular, the proposed protocol can update isolated nodes (or clusters) with missing information while ensuring homogeneous information dispersal at low overhead. Through a number of analyses, we show that SERVUS is highly efficient with regard to the tradeoff between reliability and cost, and overcomes important issues like the broadcast storm problem and the topological temporal fragility.

Magnet: a content centric routing protocol for large scale networks

Content routing is aimed to provide location-independent access to an object without relying on the maintenance of network connections between the source and the destination of the object. This paradigm enables the usage patterns in terms of applications involving data objects retrieval or services access by giving emphasis to the content rather than the location. Messages are routed on the basis of their descriptions rather than host locations. In this paper, we propose a novel content routing protocol designed for retrieving data objects in large-scale networks. Our routing scheme is based on two algorithms: request filtering and content dissemination. The design of these algorithms allows our protocol being more adapted to large scale networks than the classical content routing protocols. Our solution is based on a set of metrics and design principles aimed to achieve a compromise between the matching efficiency and the cost of communication and storage. Through conducted simulations, we show that our protocol is successful in achieving this compromise necessary for large-scale networks.

Offloading Massive Data Onto Passenger Vehicles: Topology Simplification and Traffic Assignment

Offloading is a promising technique for alleviating the ever-growing traffic load from infrastructure-based networks such as the Internet. Offloading consists of using alternative methods of transmission as a cost-effective solution for network operators to extend their transport capacity. In this paper, we advocate the use of conventional vehicles equipped with storage devices as data carriers whilst being driven for daily routine journeys. The road network can be turned into a large-capacity transmission system to offload bulk transfers of delay-tolerant data from the Internet. One of the challenges we address is assigning data to flows of vehicles while coping with the complexity of the road network. We propose an embedding algorithm that computes an offloading overlay where each logical link spans over multiple stretches of road from the underlying road infrastructure. We then formulate the data transfer assignment problem as a novel linear programming model we solve to determine the optimal logical paths matching the performance requirements of a data transfer. We evaluate our road traffic allocation scheme using actual road traffic counts in France. The numerical results show that 20% of vehicles in circulation in France equipped with only one Terabyte of storage can offload Petabyte transfers in a week.

Vehicles as Big Data Carriers: Road Map Space Reduction and Efficient Data Assignment

We advocate the use of a data shuttle service model to offload bulk transfers of delay-tolerant data from the Internet onto standard vehicles equipped with data storage capabilities. We first propose an embedding algorithm that computes an offloading overlay on top of the road infrastructure. The goal is to simplify the representation of the road infrastructure as raw maps are too complex to handle. In this overlay, each logical link maps multiple stretches of road from the underlying road infrastructure. We formulate then the data transfer assignment problem as a novel linear programming model that determines the most appropriate logical paths in the offloading overlay for a data transfer request. We evaluate our proposal using actual road traffic counts in France. Numerical results show that we can satisfy weekly aggregate requests in the petabyte range while achieving cumulative bandwidth above 10 Gbps with a market share of 20% and only one terabyte of storage per vehicle.

A data-driven analysis of YouTube community features

The success of YouTube has profoundly changed the face of industries dealing with digital content as it provides new means of distribution and promotion. While YouTube poses new opportunities for content creators to quickly reach a large audience of viewers, all videos posted online do not compete on the same footing with regard to popularity. To better understand the variation in the popularity of videos, we investigate the role of social interactions between users. In this way, our work is in stark contrast to prior research that studied user generated content video systems but without considering the structure of social relationships within those systems. In this paper, we conduct measurements on YouTube by applying a novel methodology to identify all the users interacting within the same community of interest. Using user information and the meta-data of posted videos, we analyze the influence of the community-based features of YouTube on the popularity of content posted online. Our analysis shows that users posting videos under a specific category get a better recognition than those actively posting videos belonging to a large variety of categories.

Content Centricity in Constrained Cellular-Assisted D2D Communications

The huge increase of mobile traffic in the latest years has put cellular networks under pressure. To face this situation, operators propose to adopt data offloading techniques based on device-to-device communications to alleviate their infrastructure. In this paper, we consider a specific scenario in which the cellular channel has severe capacity limitations. Existing offloading techniques focus on the underlying communication mechanisms and fail to properly manage the interest users have in content. The straightforward approach to tackle this issue is to rely on the content-centric networking (CCN) paradigm. Nevertheless, the hybrid nature of our scenario makes this vision challenging—what should circulate through the cellular channel and what should remain within the opportunistic network? In this paper, we investigate our target scenario and identify a number of challenges therein. We finally define a high-level architecture that we intend to instantiate in the case of a public infrastructure scenario.

Centrally Controlled Mass Data Offloading Using Vehicular Traffic

With over 300 billion vehicle trips made in the United States and 64 billion in France per year, network operators have the opportunity to utilize the existing road and highway network as an alternative data network to offload large amounts of delay-tolerant traffic. To enable the road network as a large-capacity transmission system, we exploit the existing mobility of vehicles equipped with wireless and storage capacities together with a collection of offloading spots. An offloading spot is a data storage equipment located where vehicles usually park. Data is transloaded from a conventional data network to the closest offloading spot and then shipped by vehicles along their line of travel. The subsequent offloading spots act as data relay boxes where vehicles can drop off data for later pick-up by other vehicles, depending on their direction of travel. The main challenges of this offloading system are how to compute the road path matching the performance requirements of a data transfer and how to configure the sequence of offloading spots involved in the transfer. We propose a scalable and adaptive centralized architecture built on software-defined networking that maximizes the utilization of the flow of vehicles connecting consecutive offloading spots. We simulate the performance of our system using real roads traffic counts for France. Results show that the centralized controlled offloading architecture can achieve an efficient and fair allocation of concurrent data transfers between major cities in France.