Anne Fladenmuller

SDN Based Architecture to Improve Video Streaming in Home Networks

Nowadays, Internet video is the dominant internet traffic. DASH is an adaptive video streaming technique introduced to enable high quality video delivery over HTTP. In home networks, multiple video streams will compete for bandwidth, thus leading to poor performance and impacting the received quality of experience. In this paper we introduce a new technique to address this issue at the home network gateway without modifying neither the client player nor the video server. We design our framework NAVS (Network Assisted Video Streaming) relies on the deployment of Software Defined Networking (SDN). NAVS performs a dynamic traffic shaping based on the collected network traffic statistics and monitoring of video flows. NAVS dynamically allocates bandwidth for each video flow in real time. NAVS scheme has been evaluated over several metrics: bandwidth utilization, instability of players as well as the average video quality received by the clients. Our results demonstrate an improvement for all these parameters.

Price: Hybrid geographic and co-based forwarding in delay-tolerant networks

Similarity-based techniques relying on contact patterns among nodes have shown to provide high delivery success rate when performing routing in delay-tolerant networks (DTN). Nevertheless, such approaches work well only when nodes exhibit high similarity values. In this paper, we propose Price (Periodicity-based Routing in Intermittently-Connected Environments), an agenda-based greedy forwarding scheme for DTNs that combines geographic-oriented forwarding with contact-based forwarding by relying packets according to predictable patterns of locations and contacts. We define the operation of Price and evaluate the performance of geographic forwarding under the framework of graph navigation to unveil a small-world phenomenon. We then gauge the performance of Price using a real-life mobility data set according to various well-known routing solutions such as Flooding, Prophet, Bubble Rap, and Wait. Simulation results show that Price largely outperforms other solutions, especially in terms of combined cost/delivery ratio.

Towards self-characterization of user mobility patterns

In this paper, we investigate the individuality of mobility patterns in wireless networks composed of IEEE 802.11 access points. We propose a mobility-aware clustering algorithm that uses roaming events as the metric to evaluate the proximity to access-points (APs) without using any geographical information. The contributions of this clustering algorithm are threefold. First, it provides a sanitized image of the topological mobility of individuals. Second, it categorizes clusters as belonging to agglomerations (where individuals pauses) or to paths between agglomerations. Third, it proposes a classification of user mobility with regard to the number of places with social meanings for the individual according to the number of visited clusters. We analyze data collected within periods ranging up to 8 months and show that the differences in activeness, coverage of mobility, home locality, and size of the list of guest locations are clearly individual-related. Such results serve as a basis for the definition of future user-centric communication systems.

Near-zero triangular location through time-slotted mobility prediction

To setup efficient wireless mesh networks, it is fundamental to limit the overhead needed to localize a mobile user. A promising approach is to rely on a rendezvous-based location system where the current location of a mobile node is stored at specific nodes called locators. Nevertheless, such a solution has a drawback, which happens when the locator is far from the source–destination shortest path. This results in a triangular location problem and consequently in increased overhead of signaling messages. One solution to prevent this problem would be to place the locator as close as possible to the mobile node. This requires however to predict the mobile node’s location at all times. To obtain such information, we define a mobility prediction model (an agenda) that, for each node, specifies the mesh router that is likely to be the closest to the mobile node at specific time periods. The location service that we propose formalizes the integration of the agenda with the management of location servers in a coherent and self-organized fashion. To evaluate the performance of our system compared to traditional approaches, we use two real-life mobility datasets of Wi-Fi devices in the Dartmouth campus and Taxicabs in the bay area of San Francisco. We show that our strategy significantly outperforms traditional solutions; we obtain gains ranging from 39 to 72% compared to the centralized scheme and more than 35% compared to a traditional rendezvous-based solution.

An SDN approach to adaptive Video Streaming in Wireless home networks

Adaptive video streaming techniques were introduced to enable high quality video delivery over HTTP. These schemes propose to choose dynamically the appropriate video rate to match the operating conditions. In home networks, wireless access is the predominant Internet access. Multiple clients/players with different link qualities compete over a limited wireless bandwidth to transfer their video streams. As a result, some users undergo unpredictable degradations of their Quality of Experience (QoE) while others benefit from these perturbations. In this paper we introduce a new technique to address this issue at the gateway without modifying neither the client nor the video server side. We design a framework WNAVS (Wireless Network Assisted Video Streaming) that relies on the deployment of Software Defined Networking (SDN). WNAVS performs a dynamic traffic shaping based on collected network traffic statistics and allocates bandwidth for the clients in real time. We evaluate WNAVS over several metrics: fairness, instability, average video quality as well as the video traffic utilization. Our results demonstrate an improvement for all these parameters.

Accurate 2-D localization of RFID tags using antenna transmission power control

Localization of tagged objects with sufficient precision is a main issue in many industrial applications. In this paper, a new approach is proposed for the localization of UHF passive tags simply using the environment learning approach. This method uses, for a given passive tag located in a 2-D space, an aggregate function of the Received Signal Strength Indicator (RSSI) for all the possible RFID readers transmission powers. Based on these measurements, we define a location signature, which is compared with those of other tags located at known positions in the neighborhood. We then implement a method based on the k-Nearest Neighbor (k-NN) algorithm to estimate the position of the target tag. Using a realistic test case involving seventy tags and four antennas, we show a very significant improvement of the localization accuracy in comparison with the results obtained using a single RSSI value.

Towards distributed geolocation for large scale disaster management

In the aftermath of major disasters such as earthquakes, locating individuals is crucial for passing on vital information, for example warnings and safety announcements. However, large scale disasters cause extensive damage to the network infrastructures and a generalized loss of communications in the chaos that ensues. This position paper presents a preliminary study for a geolocation service that relies on inter-device connections: mobile devices exchange positions of previously encountered devices when they come into contact. Every device thus builds a partial map of device locations and can use it to enforce geographic routing protocols that are resilient to large scale disasters.

Towards localization of RFID tags based on experimental analysis of RSSI

RFID has been developed for many applications such as object identification, object traceability, geometric localization, etc. In order to locate an object labeled by a tag, several characteristics of the tag responses have been studied. This yielded theoretical models whose accuracy is not sufficient in practice for targeted applications. In this paper, we investigate how to localize a tag based on the set of received signal strength indicator (RSSI) values measured for a large range of transmission powers. Using an extensive measurement campaign, we establish that the RSSI is a linear function of the transmission power of the RFID reader. Moreover, a minimum transmission power is necessary for activating the tag which also depends on its localization. As a result, the statistics of the RSSI for different transmission powers may be synthetized through a triple. We then show how this triple may be exploited for localizing a tag.

Variable-threshold buffer based adaptation for DASH mobile Video Streaming

Dynamic Adaptive Streaming over HTTP (DASH) was introduced to enable high video quality streaming over HTTP. DASH depends on the adaptation logic at the client to choose which video bitrate to stream from the content server for each chunk. For clients receiving video over a cellular network, the cellular first hop tends to be the bandwidth bottleneck and can exhibit significant swings in available bandwidth. In this paper we develop and evaluate a Dynamic Adaptation for mobile Video Streaming (DAVS), a technique that can be used within DASH adaptation to handle the significant bandwidth variability experienced by cellular mobile clients. In our scheme the main innovation is that the client chooses a bitrate based on whether the playout buffer occupancy (BO) falls below or above a dynamic threshold. In addition, the scheme attempts to minimize bitrate switching by again delaying a change of video bitrate selection by a window of time - that is also dynamically determined. We evaluate the performance of DAVS over real traces collected from a mobile network operator. DAVS shows better performance over different video streaming metrics. Furthermore, It increases the QoE by a range 15% – 55% compared to benchmark algorithms.

Crowdsourcing-based architecture for post-disaster geolocation: A comparative performance evaluation

In the aftermath of a natural or industrial disaster, locating individuals is crucial. However, disasters can cause extensive damage to the network infrastructures and a generalized loss of communication among survivors. In this article, we present a network support solution that provides a post-disaster geolocation-collecting service that relies on inter mobile device connections. On top of this dynamically built network, survivors mobile devices exchange information about geolocation of others they have encountered. Such information is routed towards predefined data collection centers using either the DTN Epidemic or Spray and Wait DTN protocol. Experiments were conducted on the ONE simulator and performance evaluation results confirm the effectiveness of our proposal.