Alaeddine El Fawal

EZ-Flow: removing turbulence in IEEE 802.11 wireless mesh networks without message passing

Recent analytical and experimental work demonstrate that IEEE 802.11-based wireless mesh networks are prone to turbulence. Manifestations of such turbulence take the form of large buffer build-up at relay nodes, end-to-end delay fluctuations, and traffic congestion. In this paper, we propose and evaluate a novel, distributed flow-control mechanism to address this problem, called EZ-flow. EZ-flow is fully compatible with the IEEE 802.11 standard (i.e., it does not modify headers in packets), can be implemented using off-the-shelf hardware, and does not entail any communication overhead. EZ-flow operates by adapting the minimum congestion window parameter at each relay node, based on an estimation of the buffer occupancy at its successor node in the mesh. We show how such an estimation can be conducted passively by taking advantage of the broadcast nature of the wireless channel. Real experiments, run on a 9-node testbed deployed over 4 different buildings, show that EZ-flow effectively smoothes traffic and improves delay, throughput, and fairness performance. We further corroborate these results with a mathematical stability analysis and extensive ns-2 simulations run for different traffic workloads and network topologies.

Vulnerabilities in Epidemic Forwarding

We identify vulnerabilities in epidemic forwarding. We address broadcast applications over wireless ad-hoc networks. Epidemic forwarding employes several mechanisms such as inhibition and spread control, and each of them can be implemented using alternative methods. Thus, the existence of vulnerabilities is highly dependent on the methods used. We examine the links between them. We classify vulnerabilities into two categories: maliccious and rational. We examine the effect of the attacks according to the number of attackers and the different network settings such as density, mobility and congestion. We show that malicious attacks are hard to achieve and their impacts are scenario dependent. In contrast, rational attackers always obtain a significant benefit. The evaluation is carried out using detailed realistic simulations over networks with up to 1000 nodes. We consider static scenarios, as well as vehicular networks.

Aziala-net: deploying a scalable multi-hop wireless testbed platform for research purposes

Aziala-net is a flexible and scalable experimental testbed or wireless multi-hop networks based on simple off-the-shelf hardware that is able to adapt to various research purposes. It is composed of more than

Coexistence of Multiple HomePlug AV Logical Networks: A Measurement Based Study

50

Empirical performance evaluation of data dissemination mechanisms for spot applications

Asus wireless routers that have been adapted to either work as fixed base station or as mobile nodes. After describing the technical details of Aziala-net, we illustrate the potential of the testbed by showing two samples of works that are currently under study in the testbed. The first example focus on the use of the IEEE 802.11 MAC layer protocol for multi-hop networks and the stability problem that it faces in the case of wireless mesh networks. The second example focus on epidemic forwarding protocols and their performance in a real testbed deployment.

Self-Limiting Epidemic Forwarding

HomePlug AV (HPAV) was designed to provide high speed in-home communication with 200Mbps PHY rate, with the goal to overcome various noises over power wires and to jump from one phase to a neighboring one. A HomePlug AV Logical Networks (AVLN) is defined by cryptographic means, i.e. stations that share the same key and hear each other are in the same AVLN; however, AVLNs which coexist in a neighborhood cannot communicate but share the same physical layer. These points imply that people using HomePlug AV may share system throughput with neighbors without being aware of it. In order to assess the reality of this potential problem, we performed measurements on an experimental testbed with several AVLNs and equipment from different manufacturers. Our results are: 1. we verified that HomePlug AV stations can communicate even over physically separated wires, and thus neighboring stations in the same AVLN or in different AVLNs may share the same throughput; 2. when stations are placed in two different AVLNs, system performance is noticeably less compared to having the same stations at the same locations but in one single AVLN; 3. HPAV stations from different manufacturers do interoperate but experience heavy per-pair throughput outages. Our findings suggest that HPAV does not perform satisfactorily in large deployments. A possible solution to the problem would be to make different AVLNs quasi-orthogonal at the physical layer, perhaps using the cryptographic key to seed an OFDM hopping sequence.

Performance Analysis of Self Limiting Epidemic Forwarding

We evaluate, by measurements on a real testbed, the performance of networking services required for spot applications. We call a “spot application” an application for smartphones (such as “Ad-hoc Flash Sales”) that disseminates information in a local neighborhood through hop-by-hop wireless forwarding. The dissemination environment is challenging. It is open-ended, and highly dynamic with very limiting resource constraints. We define five success criteria that are required by a spot application to ensure a sustainable dissemination. They cover issues such as adaptability, resource conserving and co-existence with TCP applications. We evaluate a package of mechanisms and we identify how those, through their interaction, can altogether meet these criteria. This package includes flow control, forwarding factor control and buffer management. Our measurements are carried out on a realistic testbed composed of 50 wireless devices. Our metrics include spread, application rate, forwarding factor and delay. Some findings are: blind mechanisms can interact among each other to meet the success criteria; aging entails intolerable processing complexity for smartphones; and TTL-based buffer management performs poorly with large buffer size.