Application-Aware Self-Optimization of Wireless Mesh Networks with AquareYoum and DES-SERT Barbara Staehle, Florian Wamser, Oliver Hahm, Nicolai Schmittberger, Sebastian Deschner, Andreas Blenk, Mesut Günes Dirk Staehle Julius-Maximilians-Universität Würzburg Freie Universität Berlin
AquareYoum Application and Quality of Experience Aware Resource Management for YouTube in Wireless Mesh Networks Cisco Visual Networking Index 2010 Increasing heterogeneity on application layer Increasingly different application specific QoS requirements Application and QoE Aware RM necessary for ensuring user satisfaction YouTube and WMNs as an example Application-Aware Self-Optimization of Wireless Mesh Networks 2 with AquareYoum and DES-SERT
QoE-based RRM Main idea QoE assessment at the edge of the network QoE based RRM decision in the network Utilize QoE for RM Decisions Advantage Dynamic resource adaptation to application Network quality feedback Measure QoE Measure Problem Report QoE QoE Report QoE QoE prediction necessary for avoiding QoE degradations Solution: Monitor application comfort AC = application usage and condition allows to predict QoE degradation Application-Aware Self-Optimization of Wireless Mesh Networks 3 with AquareYoum and DES-SERT
The AquareYoum Architecture Obtains information on network resources and AC Monitors available Decides upon resources application specific Sends current resource RM Tool status to mesh advisor Mesh Mesh Monitor Monitor Utilization Resource Management AC = Application usage Utilization Tools and condition Channel assignment Allows QoE prediction Traffic shaping Mesh YouTube AC = buffered Advisor Dynamic gateway selection playtime Routing path selection … Application Comfort Application Comfort Application Application Detects application Monitor Monitor Monitors AC Sends current AC to mesh advisor Application-Aware Self-Optimization of Wireless Mesh Networks 4 with AquareYoum and DES-SERT
DES-Testbed Properties of the DES-Testbed Persistent testbed Indoor nodes Outdoor nodes Mobile nodes >100 hybrid nodes Spanning ≥3 buildings Wireless mesh nodes Wireless sensor nodes Application-Aware Self-Optimization of Wireless Mesh Networks 5 with AquareYoum and DES-SERT
DES-SERT Distributed Embedded Systems - Simple and Extensible Routing Framework for Testbeds • ANSI-C library • Underlay routing in user space • Routing based on IP or MAC addresses • Multiplexing and piggybacking of routing information Periodic task execution • Runtime configuration and monitoring • • Command line interface • SNMP via AgentX Logging (file, syslog, ring buffer, …) • • Monitor interface support Application-Aware Self-Optimization of Wireless Mesh Networks 6 with AquareYoum and DES-SERT
Interaction of Components D RM Application Mesh Advisor Tool Monitor Dory E Forrest YoMo S Routing path status Bandwidth of YouTube β too - Routing path too small Alert! path small? S Monitor Routing path status buffered no yes playtime E β rating X = best position for R higher Move to X and bandwidth provide Move to position X Routing path status hop additional hop T Mobile mesh Routing path Decision about YouTube buffered playtime β monitoring node quality RM actions Application-Aware Self-Optimization of Wireless Mesh Networks 7 with AquareYoum and DES-SERT
Demo Setup A node failure 1 happens. 3 2 The routing protocol 2 replaces the broken route. 4 1 6 The bandwidth of the new route is not 3 sufficient, β falls below the alarm threshold, YoMo alarms Dory. 5 Dory triggers 4 Forrest to move. Forrest moves to the place Client displaying a YouTube video of the broken node to 5 provide an additional hop. The AquareYoum Architecture supports the self- healing properties of the DES-SERT framework The routing protocol moves 6 The two ideas complement one another to allow the flow to the new route for a successful YouTube playback Application-Aware Self-Optimization of Wireless Mesh Networks 8 with AquareYoum and DES-SERT
Recommend
More recommend
