SILENCE BASED ADAPTATION ON MULTIHOP WIRELESS NETWORKSOUTLINESImportance of Adaptivity in NetworksAudio Adaptation mechanismImportance of silence based adaptation schemesSimulationEffective BW increasingSlide 8Server consumed BW DropBRANDY MODELCharacteristics of BRANDY modelStatechart overviewAdaptation ApproachAdvantagesFuture workSILENCE BASED ADAPTATION ON MULTIHOP WIRELESS NETWORKSPALE SPANOS And FOTIOS KONSTANTINIDISOUTLINES•Importance of adaptivity in networks•Importance of silence detection and “removal”•Our Simulation•Results•Silence detection model used•Future Work based on thatImportance of Adaptivity in Networks•The nature of Ad-hoc,multihop wireless multimedia networks is that which demands adaptive mechanisms at all design layers in order to gain a desired responsiveness due to unexpected changes in grades of service•This nature consisted of the media high probability,burstiness and persistence of errors.•So,we have to adapt the operation of the applications to changing QoS•The normal Adaptive mechanism which is used is based on a periodical end-to-end feedback.Audio Adaptation mechanism•The conventional Adaptation mechanisms for audio applications are based on the QoS Notification Programming Model in which senders use QoS feedback to dynamically select audio encoding most appropriate to network conditions.•This encoding is based on the principle of media scaling.•These mechanisms are sufficient to improve the audio or video quality when lost packets are mostly caused by congestion.Importance of silence based adaptation schemes•Bandwidth requirements of video are signifficant.•Reduce the end-to-end delay seen by the users.•Imagine that the perceived audio quality drops sharply as packet loss reaches 20%,even if retransmission techniques are used to replace lost packets.Simulation•GLOMOSIM Platform•Multihop Network•number of nodes: 30•80 by 80 unit•2 Mbps radio•40 units range(reasonable probability of the generated topology being disjoint and interference between connections)•Mobility: random waypoint •Radio Type:No Capture•Propagation Function:Free SpaceEffective BW increasing•Increase effective BW when•reducing the rate at the •server side.In mobility •these are more obvious•As we can see from the figures:•1)In 802.11(mob) 102Kbps more effective BW(340%) Audio adapt•2)In CSMA (mob)15.5Kbps more effective BW(155%)Audio adaptEffective BW increasing•As concerns the silence•adaptation,the effective BW •was improved further:•1)In 802.11(mob) 169Kbps more effective BW(530%)•2)In CSMA(mob.)18Kbps more effective BW(180%)Server consumed BW Drop•In 802.11 66Kbps drop for•only Audio Adaptation(26%)•plus Silence 76.8 Kbps (30%)•In CSMA 122.8 Kbps drop for•Audio only(47%)•plus Silence 157.8 Kbps drop (62%)BRANDY MODEL•Silence in speech can be classified into 3 categories:•1)Listening pauses•2)Short speaking pauses(words,syllables)•3)Long speaking pauses(phrases,sentences)•We care for the 3rd category.Characteristics of BRANDY model•Is based on state per state probabilities.•According to these,the overall length of a talkspurt is characterized as f(t)=λe-λt •where λ=6.322 and t is expressed in seconds.•The beauty of this model is that it has an expression also for the length of silence• f(t)=G1ζe-ζt +G2γe-γt•where G1=0.8745,G2=0.1469,ζ=27.62,γ=1.377Statechart overviewAdaptation Approach•Adaptation SchemeSilenceSilenceAdvantages•This approach achieved signifficant drop of the loss rate•And drop of the server consumed BW•It seemed to work quite well with the AudioTool of the GLOMOSIM•No client’s involvement in the adaptation(possible lower delay)•Softer trafficFuture work •QoS parameter(notification packet when silence--2 packets when end)•More measurements in many different conditions to ensure compatibility with the Notification based schemes.•Timeout threshold issues(communicate with the other adaptation