Section: New Results

Performance analysis and networks protocols

Participants : Mohammed Amer, Thomas Begin, Anthony Busson, Éric Fleury, Yannick Leo, Isabelle Guerin Lassous, Philippe Nain, Huu Nghi Nguyen, Laurent Reynaud.

Network Softwarization

We have developed a modelling framework to analytically evaluate the performance of DPDK-based virtual switches in the context of NFV (Network Function Virtualisation) networks. In [34], we extended our previous work [82] to enable non-null switch-over times that account for a delay overhead whenever a CPU starts polling a different queue. More recently, in [35], we refined our framework to let it deal with batches of packets (i.e. several packets on the same queue are processed together) that tends to speed up the performance of the virtual switches. These works were partly funded by the French ANR REFLEXION under the “ANR-14-CE28-0019” project.

Wi-Fi optimization

Densification of Wi-Fi networks has led to the possibility for a station to choose between several access points (APs). On the other hand, the densification of APs generates interference, contention and decreases the global throughput as APs have to share a limited number of channels. Optimizing the association step between APs and stations can alleviate this problem and increase the overall throughput and fairness between stations. We have proposed an original solution to this optimization problem based on a mathematical model and introduce a local search algorithm to solve this problem through a suitable neighborhood structure. Our evaluation, based on simulations, shows that the proposed solution improves the overall throughput and the fairness of the network. We are currently working on variant of this problem where the traffic to the stations is taken into account in the model and the optimization formulation.

Caching

In [72] we focus on the LRU cache where requests for distinct contents are described by independent stationary and ergodic processes. We extend a TTL-based approximation of the cache hit probability first proposed by R. Fagin in 1977 for the independence reference model to this more general workload model. We show that under very general conditions this approximation is exact as the cache size and the number of contents go to infinity. Moreover, we establish this not only for the aggregate cache hit probability but also for every individual content. Last, we obtain a rate of convergence.

In [70] we consider the problem of allocating cache resources among multiple content providers. The cache can be partitioned into slices and each partition can be dedicated to a particular content provider, or shared among a number of them. It is assumed that each partition employs the LRU policy for managing content. We propose utility-driven partitioning, where we associate with each content provider a utility that is a function of the hit rate observed by the content provider. We consider two scenarios: i) content providers serve disjoint sets of files, ii) there is some overlap in the content served by multiple content providers. In the first case, we prove that cache partitioning outperforms cache sharing as cache size and numbers of contents served by providers go to infinity. In the second case, It can be beneficial to have separate partitions for overlapped content. In the case of two providers it is usually always beneficial to allocate a cache partition to serve all overlapped content and separate partitions to serve the non-overlapped contents of both providers. We establish conditions when this is true asymptotically but also present an example where it is not true asymptotically. We develop online algorithms that dynamically adjust partition sizes in order to maximize the overall utility and prove that they converge to optimal solutions, and through numerical evaluations we show they are effective.

Mobile networks

The development of analytical models to analyze the behavior of vehicular ad hoc networks (VANETs) is a challenging aim. Adaptive methods are suitable for many algorithms (e.g. choice of forwarding paths, dynamic resource allocation, channel control congestion) and services (e.g. provision of multimedia services, message dissemination). These adaptive algorithms help the network to maintain a desired performance level. However, this is a difficult goal to achieve, especially in VANETs due to fast position changes of the VANET nodes. Adaptive decisions should be taken according to the current conditions of the VANET. Therefore, evaluation of transient measures is required for the characterization of VANETs. In the literature, different works address the characterization and measurement of the idle (or busy) time to be used in different proposals to attain a more efficient usage of wireless network. We have developed an analytical model based on a straightforward Markov reward chain (MRC) to obtain transient measurements of the idle time of the link between two VANET nodes. We have shown that numerical results from the analytical model fit well with simulation results [20].

In another study, we have investigated the application of an adapted controlled mobility strategy on self-propelling nodes, which could efficiently provide network resource to users scattered on a designated area. We have designed a virtual force-based controlled mobility scheme (called VFPc) and evaluated its ability to be jointly used with a dual packet-forwarding and epidemic routing protocol. In particular, we have studied the possibility for end-users to achieve synchronous communications at given times of the considered scenarios. On this basis, we have studied the delay distribution for such user traffic and show the advantages of our solution compared to other packet-forwarding and packet-replication schemes, and highlighted that VFPc-enabled applications could take benefit of both schemes to yield a better user experience, despite challenging network conditions [21].