Projet de thèse en Réseaux, information et communications
Sous la direction de Dario Rossi.
Thèses en préparation à Paris Saclay , dans le cadre de Sciences et Technologies de l'Information et de la Communication , en partenariat avec LTCI - Laboratoire de Traitement et Communication de l'Information (laboratoire) et de Télécom ParisTech (établissement de préparation de la thèse) depuis le 16-02-2016 .
See the english version. This PhD thesis will be carried on in an international context, hence the candidate must be fluent in english.
Designing network-integrated mobility, storage and security via ICN
Over the last few years, production and consumption of digital goods increased significantly. Access to content, which is now becoming king , implies an explosion of Web-based (especially) video  services, often consumed but also and especially produced with smartphones and tablets. To face this, a radical change into the network architecture is expected and started in the 5G space. Mobility, storage and security have become a basic premise for every network communication and current over-the-top solutions appear to be unsatisfactory in terms of operational complexity and signaling. The need for a mobility management paradigm to apply within IP networks has previously striven a lot of efforts in research and standardization bodies (IETF, 3GPP among others), all resulting in a complex access-dependent set of mechanisms implemented via a dedicated control infrastructure. The complexity and lack of flexibility of such approaches (e.g. Mobile IP) calls today for a radically new solution dismantling traditional assumptions like tunneling and anchoring of all mobile communications into network core. Finally, storage capabilities have already been spread across the network in the last decade with the proliferation of content delivery solutions, data-centers etc, while always remaining application-layer solutions at the border of the network. The distribution and integration of storage as basic network primitive is today envisaged to reduce overall latency and to increase distributed computation at the edge of the network. As a third pillar of a new network design, security comes as a prerogative of today network communications and current tunnel-based solutions hardly coexist with the previous mobility and storage features by hiding the nature of traffic exchanged between two end-points. The Information Centric Network (ICN) paradigm features native support for mobility, security and storage functionalities within the network architecture and is a promising candidate for 5G networks. In particular, ICN relieves current limitations in mobility management by introducing a radically new approach: the communication focuses on names rather than on network addresses, hence a change in physical location does not imply a change in the data plane. Consumer mobility is naturally supported in virtue of its connectionless and pull-based transport mode and implies a simple retransmission by the consumer of Interests for not yet received Data. Producer mobility and real-time group communication are more challenging to support, depending on frequency of movements and on content lifetime. An additional advantage of ICN is the seamless support of multi-homed access in a datagram manner, which in case of network with transient connectivity, is preferable with respect to the increased complexity and coupled state of connection-oriented solutions using SCTP/Multi-path TCP. In-network storage is natively coupled with data transport in ICN: the provided content awareness enables an opportunistic re-use of buffered content for serving multiple requests for the same data or for loss recovery. In-network caching is also compatible with the radically different security model proposed by ICN, which replaces container-based security (contained standing for connection, link, tunnel) with content-based security. The joint design of ICN architectural components for mobility, storage and security is fundamental to enable the envisaged benefits in terms of either end user performance and simplicity/efficiency in network management. Design and evaluation of such building blocks is at the heart of this Ph.D. * Related work While ICN, and in particularly Content Centric Network (CCN) , have attracted significant attention from the scientific community[2-15], however benefits of ICN with respect to current technologies are still debated [7, 5]. While knowledge in the ICN community progresses, a number of simplification in the evaluation yield possibly a significant bias in the results. Such simplication happens at multiple levels. For instance, algorithmic settings are often sub-optimal (if not naive); content availability is considered to be homogeneous in space and time with very few exceptions [6, 4]. These simplifications impact the mobile producer use case: consider indeed that whenever a producer moves,it is necessary to proactively update (potentially global) content routing information ; since this incurs a non negligible overhead, a better approach would be to exploit caching via efficient content discovery mechanism [3, 13, 14]. Yet, efficient design mandates to consider the above workload specificities that are generally neglected: otherwise stated, a tailored design is needed to efficiently exploit correlation along the spatial and temporal axes. * Objectif In the context of ICN as a substrate of 5G, exploiting correlation in time and space seems a promising direction for a PhD Thesis. Space correlation can be exploited by adapting our previous content discovery work [3,14] to the 5G case (i.e., to cope with medium access contention and interference that are instead absent in the fixed access case). Exploiting time-correlation is feasible and without reinventing the wheel, it seems that seminal work of the VLDB community such as 2Q is worth adapting to the ICN context some 20 years later. Finally, in the producer mobility case, more subtle design is needed [16-20]: in this case caching should exploit side information, unrelated from the content request dynamics, but rather tied to the spatial mobility dynamics of the producer , which is expected to improve user experience [16,17] as well as relieving the 5G backhaul . For instance, using wireless signal strength to forecast connectivity changes/disruption, could prioritize caching of content produced by elements with higher disconnection probability, avoiding (or at least delaying) the need for routing table updates. * Method The candidate will carry on original research work pertaining to the design and performance evaluation of the above schemes. We advocate the use of multiple methodologies including (i) analytical techniques, to investigate specific system aspects and confirm the soundness of the design in the broadest possible settings; (ii) simulation/emulation techniques to confirm the performance under fewer but more realistic scenarios, (iii) prototyping of interesting solutions for demonstration, or smaller scale experimentation. Notice while not all methodologies will be necessarily used during the thesis, at least a subset of the most complementary (eg, i+ii, or i+iii) shall be employed to reinforce the soundness of the findings.