Zina Chkirbene

Hyper-Flatnet: A novel network architecture for data centers

Massive data centers form the core infrastructure support for the ever increasing cloud based services. Hence, the performance of data centers will have important influence on the scalability of these services. In particular, the data center network must be agile, available and scalable in order to respond to ever changing service requirements and application demands. Thus, the research community has proposed novel interconnect topologies including FatTree, DCell, BCube, in order to improve the performance of data centers. However, these topologies suffer from different issues such as poor scalability, bandwidth over subscription, or can be too expensive to construct. Motivated by these challenges, we propose and investigate a new topology called HyperFlatnet that combines the advantages of previous architectures while avoiding their drawbacks. HyperFlatnet is a cost effective and high-performance data center topology. It reduces dramatically the average path length and the network latency, it increases the aggregated bottleneck throughput and the bisection bandwidth compared to FlatNet connection network while using the same number of switches and links which makes the system more robust against link failures.

Incremental relaying effect on the outage probability of correlated sources transmission

In this paper we investigate different relaying strategies for correlated source transmission in block Rayleigh fading channels. Outage probability close form expressions are derived for both the cases where incremental decode and forward and only simple decode and forward techniques are used in the relays. We show that when using incremental technique in the principal relay, not only we gain in terms of channel resource saving but we get also a significant outage performance improvement compared to simple decode and forward system. Another finding of this paper is that if incremental relaying technique is used in the principal relay, the system reaches a diversity order of 2 with a significant power gain compared to a two links maximum ratio combining system.

VacoNet: Variable and connected architecture for data center networks

Todays data centers may contain tens of thousands of computers with progressively more specialized and expensive equipments. Thus, the research community has proposed various interconnect topologies e.g FatTree, Dcell and Bcube. However, these architectures are too complex and very expensive to construct and they suffer from high average path length (APL) and latency. Motivated by these challenges, we propose a new data center architecture called VacoNet that combines the advantages of existing architectures while avoiding their limitations. VacoNet is a reliable, high-performance, and cost effective data center topology that can improve the network performance in terms of average path length and network latency. In addition, VacoNet can reach even 50% in infrastructure cost reduction and the power consumption will be decreased with more than 50000 watt compared to all the previous architectures. Both theoretical analysis and simulation experiments are conducted to evaluate the overall performance of the proposed architecture.

Energy-efficient based on cluster selection and trust management in cooperative spectrum sensing

Cooperative spectrum sensing (CSS) has been proposed as a solution for radio spectrum resources scarcity problem. Clustering technique is introduced to increase the performance of CSS and overcome the shadowing and fading effects, also to reduce the control channel overhead for big number of cooperative users. However, Clustering is facing two major issues which are the data falsification caused by the malicious users and the increasing delay and energy consumption especially when the cluster heads is far away from the fusion center. In this paper, we propose a cluster and forward based on the trust CSS. By dividing all the secondary users into clusters and using of the trust threshold for selecting the most trusted CHs that can send their sensing decisions to the fusion center which can save more energy consumption and delay transmission, and improve the spectrum sensing performance.

ScalNet: A Novel Network Architecture for Data Centers

This paper presents the design, implementation and evaluation of ScalNet, novel network architecture for data centers. ScalNet is a cost-effective, high- performance and scalable interconnect with almost flat architecture. ScalNet is also characterized by its low average path length, high bisection bandwidth, high aggregate bottleneck throughput, and especially high scalability while using the same number of switches and links per server as FlatNet and BCube. Both theoretical analysis and simulation experiments are conducted to evaluate its overall performance in terms of scalability, average path length, aggregate bottleneck throughput, and fault tolerance.

Efficient techniques for energy saving in data center networks

Abstract Data centers are constructed with a huge number of network devices to support the expanding cloud based services. These devices are used to achieve the highest performance in case of full utilization of the network. However, the peak capacity of the network is rarely reached. Consequently, many devices are set into idle state and cause a huge energy waste leading to a non-proportionality between the network load and the energy consumed. In this paper, we propose a new approach to improve the efficiency of data centers in terms of energy consumption. Our approach exploits the correlation in time of the inter-node communication traffic and some topological features to maximize energy saving with only a minor increase in the average path length. Our approach dynamically controls the number of active communication links by turning off and on ports in the network (switches ports and nodes ports). Simulations results confirmed the energy saving gain procured by the proposed approach with a low impact on the average path length.

Integrating Variability Management in Data Center Networks

Data centers have an important role in supporting cloud computing services (i.e. checking social media, sending emails, video conferencing,..). Hence, data centers topologies design became more important and must be able to respond to ever changing service requirements and application demands. An ultimate challenge in this research is the design of data center network that interconnects the massive number of servers, and provides efficient and fault-tolerant routing algorithm. Several topologies such as DCell, FlatNet and ScalNet have been proposed. However, these topologies generally seek to improve the scalability without taking into consideration the energy usage neither the network nfrastructure cost which is critical parameters in data centers. Motivated by these challenges, we propose a new network topology for data center, called AdyNet. It is an adaptive, dynamic, cost effective and highly performing topology. While reducing largely the infrastructure cost and the energy consumption, AdyNet outperforms FlatNet and ScalNet in terms of Average Path Length.

LaCoDa: Layered connected topology for massive data centers

One of the fundamental challenges of existing data centers is to design a network that interconnects massive number of servers, and therefore providing an efficient and fault-tolerant routing service to upper-layer applications. Several solutions have been proposed (e.g. FatTree, DCell and BCube), however they either scale too fast (i.e., double exponentially) or too slow. This paper proposes a new data center topology, called LaCoDa, that combines the advantages of previous topologies while avoiding their limitations. LaCoDa uses a small node degree that matches physical restriction for servers, and it also interconnects a large number of servers while reducing the wiring complexity and without affecting the existing servers. LaCoDa can interconnect up to millions of servers with a small number of port switches by increasing the number of directly connected clusters per layer, and therefore avoiding the cluster connections repetitions. The theoretical and experimental results show that the average path length (APL) of a 1000-node LaCoDa is smaller than the APL of a 23-node FlatTree and 11-node BCube and DCell. Moreover, with 6-port switches and a node degree of 4, the bisection bandwidth of 4.19 Million-node LaCoDa is 7% and 14% bigger than the bisection bandwidth of 310-node DCell and BCube, respectively. Consequently, LaCoDa provides better performance than DCel, BCube, FatTree in terms of average path length, throughput, latency and bisection bandwidth.