Universidad Carlos III de Madrid PeertoPeer Television for

Universidad Carlos III de Madrid Peer-to-Peer Television for the IP Multimedia Subsystem Author Alex Bikfalvi Advisor Jaime García-Reinoso

Outline I Background Motivation • Contributions • Peer-to-peer streaming • IP Multimedia Subsystem II Peer-to-Peer Television for the IMS Service architecture • Signaling protocol • Support for mobility III The User Activity in IPTV Data and modeling • Synthesis IV Enhancements at the Application Server Signaling delay • Multiple TV channels V Performance Evaluation Basic evaluation • Hybrid streaming VI Conclusions Summary • Publications • Future enhancements July 18, 2012 Peer-to-Peer Television for the IP Multimedia Subsystem 2

Part I Background July 18, 2012 Peer-to-Peer Television for the IP Multimedia Subsystem 3

Motivation Internet Protocol Television 1 • Increasing interest in the recent years • Deployments and triple-play packages • Competition with Internet Services 2 3 Next Generation Network • Flexible platform for any type of service • Decouples service provisioning from the network • Adequate level of quality of service (Qo. S) IP Multimedia Subsystem • Framework for IP multimedia services • Session control using the Session Initiation Protocol (SIP) 4 July 18, 2012 TISPAN • Extended IMS NGN to multiple access technologies • Services standardization, including IPTV Peer-to-Peer Television for the IP Multimedia Subsystem 4

Motivation In TISPAN, broadcast television uses IP multicast • High performance • Availability with existing protocols and equipments • Commercial IPTV deployments: walled gardens • • • Large number of TV channels Static multicast: inefficient for many TV channels Dynamic multicast: delay and scalability issues Administrative and economic reasons Support for multiple transport protocols July 18, 2012 Peer-to-Peer Television for the IP Multimedia Subsystem 5

Going Peer-to-Peer… Peer-to-peer television (P 2 PTV) service for the IP Multimedia Subsystem • Exploit the unused download and upload capacity July 18, 2012 Peer-to-Peer Television for the IP Multimedia Subsystem 6

Going Peer-to-Peer… Peer-to-peer television (P 2 PTV) service for the IP Multimedia Subsystem • Exploit the unused download and upload capacity • Dedicated user equipments (set-top boxes) • Streaming transparent to the user July 18, 2012 Peer-to-Peer Television for the IP Multimedia Subsystem 7

… with the IMS Peer-to-peer television (P 2 PTV) service for the IP Multimedia Subsystem Application Server • Control service access • Manage peer participation • Implement enhancements User Equipment (peer) • Download current channel • Download other streams • Upload streams to other peers July 18, 2012 Peer-to-Peer Television for the IP Multimedia Subsystem 8

Challenges We don’t break new ground in peer-to-peer streaming However… • Integration with a commercial-grade service • Multiple TV channels increase peer churn (over 60% of channel changes in 10 seconds[1, 2]) • IMS signaling requirements increase the setup delay 1 Two enhancements… Fast signaling Inactive uploading sessions with committed Qo. S resources 2 Low churn Peer participation on multiple TV channels[3] [1] Cha et al. , Watching Television Over an IP Network, 2008 [2] Qiu et al. , Modeling user activities in a large IPTV system, 2009 [3] Wu et al, . View-upload decoupling: A redesign of multichannel P 2 P video systems, 2009 July 18, 2012 Peer-to-Peer Television for the IP Multimedia Subsystem 9

Part I Background Peer-to-Peer Streaming July 18, 2012 Peer-to-Peer Television for the IP Multimedia Subsystem 10

Peer-to-Peer Streaming • Emerged as a response to IP multicast issues Support IP multicast Low CDN P 2 P Scalabilit y Cost Med Low Med High Resourc es Complex Low High Med Low High • Initially P 2 P emulated IP multicast • Application-level multicast • Data forwarded along a tree overlay between hosts July 18, 2012 Peer-to-Peer Television for the IP Multimedia Subsystem 11

Peer-to-Peer Streaming Overlay tree between end-hosts or peers • Data flow may be push or pull • Accommodates one stream or channel July 18, 2012 Peer-to-Peer Television for the IP Multimedia Subsystem 12

Peer-to-Peer Streaming Overlay tree between end-hosts or peers • Data flow may be push or pull • Accommodates one stream or channel • Churn: interruptions due to the departing peers July 18, 2012 Peer-to-Peer Television for the IP Multimedia Subsystem 13

Multiple TV Channels Increased peer churn due to channel changes • View-upload decoupling[1] Wu et al. , View-upload decoupling: A redesign of multichannel P 2 P video systems, 2008 July 18, 2012 Peer-to-Peer Television for the IP Multimedia Subsystem 14

Part I Background The IP Multimedia Subsystem July 18, 2012 Peer-to-Peer Television for the IP Multimedia Subsystem 15

The IP Multimedia Subsystem Next Generation Networks Integrated broadband IP networks for multimedia services Third Generation Partnership Project (3 GPP) IP Multimedia Subsystem Decouples service from the transport network Functional entities and standardized interfaces July 18, 2012 • • Quality of service Service implementation Seamless mobility Authentication, policy and charging Peer-to-Peer Television for the IP Multimedia Subsystem 16

The IP Multimedia Subsystem Gateways Application plane Control plane Policy and charging Transport plane Proxy Serving Call-session control July 18, 2012 Peer-to-Peer Television for the IP Multimedia Subsystem Applications servers Subscriber database 17

Part II Peer-to-Peer Television for the IP Multimedia Subsystem July 18, 2012 Peer-to-Peer Television for the IP Multimedia Subsystem 18

Service Architecture Common platform for IPTV streaming using peer-topeer technology: P 2 PTV Broadcast Servers P 2 PTV Application Server User Equipment July 18, 2012 Peer-to-Peer Television for the IP Multimedia Subsystem 19

Application Server SIP Signaling Manage the IMS multimedia sessions with the UE sip: p 2 ptvas. example. net User agent server • Sessions terminating at the P 2 PTV-AS • Streaming by the broadcast server P 2 PTV-AS UAS 2 sip: alice@example. n et UE July 18, 2012 S-CSCF 1 INVITE sip: p 2 ptv@example. net From: sip: alice@example. net To: sip: p 2 ptv@example. net Peer-to-Peer Television for the IP Multimedia Subsystem Call-ID: 1000 20

Application Server Back-to-back user agent SIP Signaling • Peer-to-peer streaming Manage the IMS multimedia sessions with the UE sip: p 2 ptvas. example. net P 2 PTV-AS B 2 BUA sip: alice@example. n et UE From: sip: alice@example. net To: sip: p 2 ptv@example. net Call-ID: 3000 July 18, 2012 S-CSCF sip: bob@example. net UE From: sip: p 2 ptv@example. net To: sip: bob@example. net Call-ID: 4000 Peer-to-Peer Television for the IP Multimedia Subsystem 21

Channel Streaming The P 2 P push-pull streaming generates an overlay A TV channel is divided in multiple (e. g. 3) streams Ideally, a UE peer downloads all streams when tuning to the channel A UE peer may upload one or more streams to overlay neighbors This strategy works well with multiple description codecs such as H. 264/SVC July 18, 2012 Peer-to-Peer Television for the IP Multimedia Subsystem 22

Streaming Enhancements We face the following challenges… • The IMS session signaling is an expensive operation • Streaming multiple channels with a classic experience 1 We propose two solutions… Fast signaling Inactive uploading sessions with committed Qo. S resources 2 Low churn Peer participation on multiple TV channels • The first targets the control plane • The second targets the media plane July 18, 2012 Peer-to-Peer Television for the IP Multimedia Subsystem 23

Fast Signaling The P 2 P streaming requires two multimedia sessions • Downloading side, low dynamics, reusable during channel changes • Uploading side, high dynamics, non-reusable during channel changes The performance bottleneck is on the upload side Introduce foster peers with inactive upload sessions These accommodate new requests without establishing new sessions July 18, 2012 Peer-to-Peer Television for the IP Multimedia Subsystem 24

Fast Signaling Situations that benefit from foster peers • Fast stream change when the user changes the current channel • Fast recovery to accommodate peer churn when occurs Changing stream 1 2 Number of inactive sessions Fast stream change 1 Fast recovery July 18, 2012 Peer-to-Peer Television for the IP Multimedia Subsystem 25

Fast Signaling Does not require a new session on the upload side • Initiated by the P 2 PTV-AS as a response to user demand sip: alice@example. net P 2 PTVAS 1 sip: bob@example. ne S-CSCF INVITE P-CSCF INVITE sip: alice@example. net. . . From: sip: p 2 ptv@example. net. . . i=stream 100 a=inactive PCRF INVITE 183 Session Progress AA-Request 183 Session Progress AA-Answer July 18, 2012 UE Peer-to-Peer Television for the IP Multimedia Subsystem 2 26

Fast Signaling Changing the TV stream using a foster peer sip: alice@example. net UEdown 1 sip: bob@example. ne P-CSCF UPDATE sip: p 2 ptv@example. net. . . From: sip: alice@example. net To: sip: p 2 ptv@example. net. . . i=stream 100 c=IN IP 4 10. 0. 0. 1 a=curr: qos local recv a=curr: qos remote send 200 OK July 18, 2012 P 2 PTVAS S-CSCF UPDATE UEup UPDATE sip: bob@example. net. . . From: sip: p 2 ptv@example. net To: sip: bob@example. net. . . i=stream 100 c=IN IP 4 10. 0. 0. 1 a=curr: qos local send a=curr: qos remote recv 200 OK Peer-to-Peer Television for the IP Multimedia Subsystem 200 OK 27

Fast Signaling It sounds simple, but… How many inactive sessions accommodate the TV channel demand? • Too few, no fast signaling and high channel change delay • Too many, waste network resources with reserved bandwidth wi On a given TV channel July 18, 2012 1 User activity 2 Inactive sessions Peer-to-Peer Television for the IP Multimedia Subsystem 28

Peer Churn Peers download streams from multiple TV channels • Primary streams correspond to the current TV channel • Secondary streams from other TV channels Only the primary streams are affected by the channel change churn Primary streams are used for viewing Secondary streams are used for uploading The cost is the increased bandwidth usage July 18, 2012 Peer-to-Peer Television for the IP Multimedia Subsystem 29

Peer Churn This is view-upload decoupling: can we do better? • Complete decoupling wastes bandwidth • Upload primary streams for peers with free bandwidth Peers may need a new download session at the next channel change Primary streams become secondary Self-organizing, no centralized assignment of secondary streams Higher delay: use inactive download sessions July 18, 2012 Peer-to-Peer Television for the IP Multimedia Subsystem 30

Peer Coordination • The session coordination computes the number of inactive sessions for a channel wi The peer coordination assigns peer resources… • … to accommodate the demand We discuss both algorithms in further detail in part IV July 18, 2012 Peer-to-Peer Television for the IP Multimedia Subsystem 31

Part II Peer-to-Peer Television for the IP Multimedia Subsystem Support for Mobility July 18, 2012 Peer-to-Peer Television for the IP Multimedia Subsystem 32

Support for Mobility We examine the performance in roaming situations Minimize the loss of streaming data… • Buffering mechanism compensating for connectivity loss • Reducing the handover delay Existing solutions… 1 SIP 2 Establish a new session after roaming to the new network July 18, 2012 Optimized SIP Transfer the session context between the old and new PCSCF to meet the session preconditions Mobile IP 3 Tunnel the video data from the home to the visited network Peer-to-Peer Television for the IP Multimedia Subsystem 33

Proactive Context Transfer Unfortunately… The UE must reestablish the session in the new network • We exploit the handover delay when the UE is disconnected • The network takes an active participation in the handover • Use the IEEE 802. 21 (MIH) standard • One in every network Proactive • Part of the MIH point-of-service Context • Notified by the UE before the Transfer handover Service • Installs the session context in the Application network at the P-CSCF Server • Applies to SIP or MIP mobility July 18, 2012 Peer-to-Peer Television for the IP Multimedia Subsystem 34

Performance Evaluation Comparing the handover delay with previous scenarios • Delay components Visited Home SIP MIP SIP and MIP handover delay July 18, 2012 Total handover delay for UMTS Peer-to-Peer Television for the IP Multimedia Subsystem 35

Part III The User Activity July 18, 2012 Peer-to-Peer Television for the IP Multimedia Subsystem 36

Objectives Essential for system design and performance evaluation Measurement studies… • Internet-based services like PPLive, PPStream, Sop. Cast[1, 2, 3, 4, 5] • Telco IPTV such as Telefonica Imagenio, AT&T USimulwatch • Limited number of properties Verse[6, 7, 8] Synthetic workload • Low accuracy for some metrics generator by Qiu et • Some flaws al. [1] Ali et al. , Measurement of commercial peer-to-peer live video streaming, 2006 [2] Hei et al. , A measurement study of a large-scale P 2 P IPTV system, 2007 [3] Silverston et al. , Measuring P 2 P IPTV systems, 2007 [4] Xie et al. , A measurement of a large-scale peer-to-peer live video streaming system, 2007 [5] Vu et al. , Measurement of a large-scale overlay for multimedia streaming, 2007 [6] Cha et al. , Watching television over an IP network, 2008 [7] Qiu et al. , Modeling user activities in a large IPTV system, 2009 [8] Qiu al. , Modeling channel popularity. Television dynamics large IPTV system, 2009 July 18, et 2012 Peer-to-Peer for in thea. IP Multimedia Subsystem 37

User Activity The state of the user equipment and current channel[1] Offline session Online session Channel session [1] Qiu et al. , Modeling user activities in a large IPTV system, 2009 July 18, 2012 Peer-to-Peer Television for the IP Multimedia Subsystem 38

User Activity The state of the user equipment and current channel[1] Online events Offline events • Session length • Session rate July 18, 2012 Peer-to-Peer Television for the IP Multimedia Subsystem 39

Session Length • A hyper-exponential distribution Online session interval length • Fitting algorithm of Feldmann et al. [1] • Fit each exponential on exponentially spaced intervals [1] Feldmann et al. , Fitting mixtures of exponentials to long-tail distributions to analyze network performance models, 1998 July 18, 2012 Peer-to-Peer Television for the IP Multimedia Subsystem 40

Session Rate • Has a complex daily and weekly pattern 60 20 min 30 min 15 min 10 min 6. 6 6 mi n Difficult to model Online session rate (normalized) Frequency spectrum • Qiu et al. model the spectrum with a continuous distribution • Limited accuracy: does not include phase information • July Propose dominant frequency components based on Peer-to-Peer Television for the IP Multimedia Subsystem 18, 2012 power 41

Session Rate • Has a complex daily and weekly pattern 60 20 min 30 min 15 min 10 min 6. 6 6 mi n Difficult to model 59+57 parameters 9+7 parameters Online session rate (normalized) Frequency spectrum • Qiu et al. model the spectrum with a continuous distribution • Limited accuracy: does not include phase information • July Propose dominant frequency components based on Peer-to-Peer Television for the IP Multimedia Subsystem 18, 2012 power 42

Session Rate • Include weekly pattern using a modulating function • Based on the number of online viewers • Stochastic properties • Difference between trace and model: normal distribution July 18, 2012 Peer-to-Peer Television for the IP Multimedia Subsystem 43

Workload Synthesis • Generate workload based on analytical model • Incomplete measurement data on IPTV user activity[1, 2, 3] • Rescale workload dimensions like users or channels Offline event Online interval Timeline Channel event Offline interval Channel interval • Conclusions • Better approximation of the activity data • Exclude some details like user preference [1] Cha et al. , Watching television over an IP network, 2008 [2] Qiu et al. , Modeling user activities in a large IPTV system, 2009 [3] Qiu et al. , Modeling channel popularity dynamics in a large IPTV system, 2009 July 18, 2012 Peer-to-Peer Television for the IP Multimedia Subsystem 44

Part IV The Application Server July 18, 2012 Peer-to-Peer Television for the IP Multimedia Subsystem 45

Application Server Functions 1 SIP Signaling Manage the IMS multimedia sessions with the UE 2 Session Coordination Compute the number of inactive upload sessions 3 Peer Coordination Assignment of peer bandwidth across TV streams July 18, 2012 Peer-to-Peer Television for the IP Multimedia Subsystem 46

Session Coordination • The system is like a queue for every TV channel Disturbance User activity u(t), zch(t) On a given TV channel Arrival zch Input Blocking ratio β and utilization ρ July 18, 2012 Blocking Active sessions wa Inactive sessions wi From the perspective of Service dch the fast signaling. Peer-to-Peer Television for the IP Multimedia Subsystem Output Number of sessions w(t)=wa(t)+wi(t) 47

Session Coordination • Finding a relationship between input and output • No simple distribution for the user activity dynamics • The temporal dimension is an important element Arrivals blocked If the arrival would be a Poisson process and the service rate have an exponential distribution The number of upload session computed using the Erlang-B equation No strong correlation between the channel user activity and the optimal number of upload sessions Arrivals Simulation of the user arrival served July 18, 2012 Peer-to-Peer Television for the IP Multimedia Subsystem 48

Session Coordination Use an adaptive algorithm with a feedback loop Input: r Time-discrete system: t k Output: w(k) SCA User activity: u(k) Reference: r Input: w(k) Error: e(k) Controller Design tasks… Output: y(k) System Delay • Selection of the control signals: input, output and reference • Determine the controller transfer function: July 18, 2012 Peer-to-Peer Television for the IP Multimedia Subsystem 49

Session Coordination • Performance evaluation • Blocking A largeratio P 2 PTV deployment with 100 000 subscribers around the The session utilization is desired reference value spanning 1 week between 90% and 100% • Synthetic workload Blocking ratio b(k) July 18, 2012 Session utilization ρ(k) Peer-to-Peer Television for the IP Multimedia Subsystem 50

Peer Coordination Based on the view-upload decoupling idea…[1] Download Upload Primary streams: np Secondary streams: ns Free download bandwidth Active streams: na Inactive streams: ni Free upload bandwidth • Assign peers to secondary and inactive streams • Subject to peer bandwidth constraints We leverage the session coordination to estimate the channel demand allocate peer resources [1] Wu et al. , View-upload decoupling: A redesign of multichannel P 2 P video systems, 2008 July 18, 2012 Peer-to-Peer Television for the IP Multimedia Subsystem 51

Peer Coordination • Peers self-organize on secondary streams based on On the download side • Requests for active and inactive sessions • Available bandwidth Primary Secondary Free 1 4 10 11 12 13 20 30 40 Initial peer state 2 5 10 11 12 13 20 30 40 Receive request to upload stream 13 3 5 10 11 12 13 4 July 18, 2012 4 50 51 52 53 20 30 40 13 20 30 40 Reserve bandwidth for one additional primary stream At the next channel change the primary stream becomes secondary Peer-to-Peer Television for the IP Multimedia Subsystem 52

Peer Coordination • Key differences to view-upload decoupling Channel changes Primary Secondary View-upload decoupling Uploading primary Peer coordination algorithm • Peers report their uploading capabilities • The algorithm uses a resource level metric r to select peerslooking up When peers with free bandwidth July 18, 2012 peers with inactive sessions Peer-to-Peer Television for the IP Multimedia Subsystem 53

Peer Coordination • Performance evaluation • Three bandwidth scenarios based on DSL access Fraction Poor Middle-class Rich Download Upload 15 % 15 000 544 23 576 3 328 20 000 10 000 20 % 13 000 800 21 576 2 944 20 000 10 000 50 % 10 000 2 584 18 576 2 548 20 000 15 % 12 000 2 584 14 576 2 548 20 000 10 000 ADSL 2 and ADSL 2+ July 18, 2012 ADSL 2+ Peer-to-Peer Television for the IP Multimedia Subsystem Fiber or other broadband 54

Peer Coordination • Impact of the peer bandwidth Poor peers may not support many neighbors Always free bandwidth Middle-class close to rich Free bandwidth only for rich Committed peer bandwidth Free peer bandwidth • Uplink constrained for poor and middle-class • Increased download for poor peers • Middle-class close to necessary bandwidth July 18, 2012 Peer-to-Peer Television for the IP Multimedia Subsystem 55

Peer Coordination • Streaming overhead 1 Overhead of the secondary streams Secondary streams 2 Inactive sessions Server upload Peer download and server Inactive sessions Peer upload • Overhead diminishes for rich peers • Server contribution significant only for resource limited • Inactive sessions use a small fraction July 18, 2012 Peer-to-Peer Television for the IP Multimedia Subsystem 56

Part V Performance Evaluation July 18, 2012 Peer-to-Peer Television for the IP Multimedia Subsystem 57

Experimental Setting Performance evaluation using computer simulations Scalability Planet Lab Complexity Accuracy Reproducibilit y Medium High Emulation Low Medium High Simulation High Low • Large-scale measurements with up 100 000 viewers • We implemented two distinct delay models[1, 2, 3] • Deterministic and queuing [1] Pesch et al. , Performance evaluation of SIP-based multimedia services in UMTS, 2005 [2] Ulvan et al. , Analysis of Session Establishment Signaling Delay in IP Multimedia Subsystem, 2009 [3] Munir, Analysis of SIP-based IMS session establishment signaling for Wi. Max-3 G networks, 2008 July 18, 2012 Peer-to-Peer Television for the IP Multimedia Subsystem 58

Delay and Churn • System performance from the end-user perspective The remaining due to users turning off their UE Less than 700 ms Channel connection delay Churn performance • Depends on model of signaling processing at IMS functions [1] Kooij et al. , Perceived quality of channel zapping, 2005 [1] Peer-to-Peer Television for the IP Multimedia viewer Subsystem experience July 18, 59 • 2012 Below the threshold of acceptable

Hybrid Streaming Bandwidth poor scenarios are difficult for P 2 PTV • Peer uploading bandwidth is a limiting factor • Broadcast server participation remains significant Use P 2 P in conjunction with IP multicast? July 18, 2012 • Scalability with the number of TV channels • IP multicast for popular TV channels • P 2 P for unpopular TV channels Peer-to-Peer Television for the IP Multimedia Subsystem 60

Hybrid Streaming • Comparison of bandwidth usage Always overhead in the access Benefit for popular channels Use P 2 P for unpopular channels Core network bandwidth Access network bandwidth • Use IP multicast for most popular channels • Use peer-to-peer for the majority of unpopular channels • Always have some overhead due to secondary streams July 18, 2012 Peer-to-Peer Television for the IP Multimedia Subsystem 61

Hybrid Streaming • Comparison of scalability issues Multicast routing entries Multicast entries Server usage in rich scenarios Gain limited for unpopular Server and total bandwidth channels • Number of routing entries approximated as July 18, 2012 Peer-to-Peer Television for the IP Multimedia Subsystem 62

Conclusions July 18, 2012 Peer-to-Peer Television for the IP Multimedia Subsystem 63

Summary of Contributions Service architecture 1 2 Multi-layer streaming 3 Mobility support Workload generator 4 5 User activity model 6 Synthesis algorithms Enhancements 7 10 8 Fast signaling 9 Bandwidth assignment Performance evaluation 11 Basic concept 12 Extended scenarios July 18, 2012 Bikfalvi et al. , Nozzilla: A Peer-to-Peer IPTV Distribution Service for an IMS-based NGN, ICNS, 2009 Bikfalvi et al. , A Peer-to-Peer IPTV Service Architecture for the IP Multimedia Subsystem, IJCS, 2009 Vidal et al. , Enabling Layered Video Coding for IMS-based IPTV Home Services, IEEE Network, 2009 Vidal et al. , Supporting Mobility in an IMSbased P 2 P IPTV Service, Computer Communications, 2010 Bikfalvi et al. , P 2 P vs. IP Multicast: Comparing Approaches to IPTV Streaming Based on TV Channel Popularity, Computer Networks, 2011 Submissions in progress Peer-to-Peer Television for the IP Multimedia Subsystem 64

Future Enhancements 3 2 1 Extending the Experimental Evaluation • Assessment of playback-quality and viewer experience • Consider packet-based flows • Include codec characteristics Enhancing the Peer Coordination • Peer churn when viewers turn off their equipment • Estimate the reliability of user connection • Use this information during the peer selection Enhancing the Session Coordination • Peer churn not included in evaluation of inactive sessions • Use a feedback loop over the peer coordination • Superior blocking ratio performance July 18, 2012 Peer-to-Peer Television for the IP Multimedia Subsystem 65

Q&A

Thank You

Backup Slides July 18, 2012 Peer-to-Peer Television for the IP Multimedia Subsystem 68

Summary • The main topics covered by this dissertation Internet Protocol Television IP Multimedia Subsystem • Audio/video streaming • Commercial grade service • Open platform vs. walled gardens • Convergence of services • Guaranteed quality of service • Session signaling using SIP Peer-to-Peer Enhancement Algorithms • Uploading by user equipments • Inherent scalability • Complexity and churn • Fast signaling • Reduced peer churn • Support for mobility July 18, 2012 Peer-to-Peer Television for the IP Multimedia Subsystem 69

Multiple Trees • Addresses the issues of single trees[1] • Improves peer participation • Increases robustness to peer churn [1] Castro et al. , Split. Stream: high-bandwidth multicast in cooperative environments, 2003 July 18, 2012 Peer-to-Peer Television for the IP Multimedia Subsystem 70

Data-Driven Streaming • Focus on video streaming • Missing video pieces (segments, chunks) • Heterogeneous bandwidth and delay • Mesh overlay • Unstructured protocol July 18, 2012 Peer-to-Peer Television for the IP Multimedia Subsystem 71

Walled Gardens Commercial IPTV offering a classic viewing experience • Static IP multicast for all TV channels, unsustainable for future growth[1] Cha et al. , On next-generation telco -managed P 2 P TV architectures, 2008 July 18, 2012 Peer-to-Peer Television for the IP Multimedia Subsystem 72

Detailed Architecture P 2 PTV-AS – BS interface DSL access network July 18, 2012 UE – IMS interface (Gm) Peer-to-Peer Television for the IP Multimedia Subsystem UMTS access network 73

Application Server SIP Signaling User agent server • Sessions terminating at the P 2 PTV-AS • Streaming by the broadcast server Manage the IMS multimedia sessions with the UE sip: p 2 ptvas. example. net P 2 PTV-AS UAS 3 sip: alice@example. n et UE July 18, 2012 S-CSCF 1 2 User Service Profile Initial Filter Criteria Filter Criterion Trigger Point sip: p 2 ptv@example. Request = net URI Application Server sip: p 2 ptv-as. example. net INVITE Filter Criterion sip: p 2 ptv@example. net From: sip: alice@example. net To: sip: p 2 ptv@example. net Peer-to-Peer Television for the IP Multimedia Subsystem Call-ID: 1000 74

Application Server 1 User agent client SIP Signaling • Sessions originating at the P 2 PTV-AS • Inactive uploading sessions Manage the IMS multimedia sessions with the UE sip: p 2 ptvas. example. net P 2 PTV-AS UAC S-CSCF sip: bob@example. net UE From: sip: p 2 ptv@example. net To: sip: bob@example. net Call-ID: 2000 July 18, 2012 Peer-to-Peer Television for the IP Multimedia Subsystem 75

Session Establishment Multimedia sessions with the Session Initiation Protocol sip: alice@example. net UE 1 sip: as. example. net P-CSCF INVITE S-CSCF AS INVITE sip: service@example. net. . . From: sip: alice@example. net 2 Evaluation of the Initial Filter Criteria INVITE 3 Service control 183 Session Progress 4 July 18, 2012 Authorize resources Peer-to-Peer Television for the IP Multimedia Subsystem 76

Session Establishment Multimedia sessions with the Session Initiation Protocol sip: alice@example. net sip: as. example. net UE P-CSCF PRACK S-CSCF PRACK 200 OK 200 OK UPDATE 200 OK ACK ACK 5 Resource reservation 6 AS Content Server Multimedia content July 18, 2012 Peer-to-Peer Television for the IP Multimedia Subsystem 77

Session Signaling • The UE peers maintain a multimedia session for each downloaded or uploaded stream sip: alice@example. net UEdown 1 sip: bob@example. ne P-CSCF INVITE P 2 PTVAS S-CSCF INVITE sip: p 2 ptv@example. net. . . From: sip: alice@example. net. . . i=stream 100 2 Evaluation of the Initial Filter Criteria INVITE 3 Selection of the uploading peer INVITE sip: bob@example. net. . . From: sip: p 2 ptv@example. net July 18, 2012 UEup Peer-to-Peer Television for the IP Multimedia Subsystem INVITE 78

Session Signaling sip: alice@example. net UEdown sip: bob@example. ne P-CSCF P 2 PTVAS S-CSCF UEup 183 Session Progress 5 Authorize resources PRACK 6 Resource reservation 200 OK PRACK 200 OK 7 Resource reservation Stream 100 July 18, 2012 Peer-to-Peer Television for the IP Multimedia Subsystem 79

Business Model • The P 2 PTV relies on unused network capacity • Bandwidth available in the access network • But not contracted by the customer Requires an agreement between the P 2 PTV and the transport provider (telco) Multiple IPTV content providers may use the P 2 P streaming infrastructure An ADSL example July 18, 2012 Peer-to-Peer Television for the IP Multimedia Subsystem 80

Business Model The NGN business model facilitates the convergence of multiple services Customers… • Have a contract with a service packager (telco) • Service providers establish relationships with the transport provider July 18, 2012 Peer-to-Peer Television for the IP Multimedia Subsystem 81

Streaming Architecture We propose a push-pull mechanism • A trade-off between data-driven and session requirements 1 Pull: request streaming from segment 109 2 Push: send segments in order • Prevents playback gaps and interruptions • Helps synchronizing multiple streams of the same TV channel 18, 2012 Peer-to-Peer Television for the IP Multimedia 82 • July Streaming boost for buffer under runs. Subsystem

P 2 PTV for Mobile Devices • A mobile user equipment in an UMTS access network • Mobile UEs do not upload content due to limited bandwidth July 18, 2012 Peer-to-Peer Television for the IP Multimedia Subsystem 83

Proactive Context Transfer Obtaining the destination configuration Contact Po. S to obtain the information 2 on the surrounding networks Upon the interaction with the Po. S 3 from the surrounding networks 1 July 18, 2012 Contact MIIS to obtain the surrounding networks Peer-to-Peer Television for the IP Multimedia Subsystem 84

Session Rate • Include weekly pattern using a modulating function • Based on the number of online viewers • The new session rate is • Where we have • Stochastic properties • Difference between trace and model: normal distribution July 18, 2012 Peer-to-Peer Television for the IP Multimedia Subsystem 85

Channel Popularity • Selection of the TV channel at a channel change[1] Channel selection Target : 44 % Sequential : 56 % Turn-on : 4% Browsing : 52 % Browsing forward Browsing backward : 37 % : 15 • Popularity model for target switching • Mean: zipf/exponential distribution • Instantaneous: mean reversion model [1] Qiu et al. , Modeling channel popularity dynamics in a large IPTV system, 2009 July 18, 2012 Peer-to-Peer Television for the IP Multimedia Subsystem 86

Workload Evaluation • Session rate : Deviation due to the weekly modulating function Online session rate Offline session rate • Keeps the dominant low and high frequency components • Adds the weekly modulation and stochastic properties July 18, 2012 Peer-to-Peer Television for the IP Multimedia Subsystem 87

Workload Evaluation • Session rate : Very high frequency components are lost Channel session rate • Keeps the dominant low and high frequency components • Adds the weekly modulation and stochastic properties July 18, 2012 Peer-to-Peer Television for the IP Multimedia Subsystem 88

Workload Evaluation Error due to the trace • Sessionmismatch length : the session with rate Online session length Error due to the workload generation algorithm Offline session length • Follows closely the model probability distribution • Errors due to limited data and algorithm approximations July 18, 2012 Peer-to-Peer Television for the IP Multimedia Subsystem 89

Workload Evaluation • Session length : Channel session length • Follows closely the model probability distribution • Errors due to limited data and algorithm approximations July 18, 2012 Peer-to-Peer Television for the IP Multimedia Subsystem 90

Workload Evaluation • Number of online viewers : Peaks lost due to missing of high frequency rates Some days approximated well Some days approximated poorly Number of online viewers (normalized) • Number of online viewers determined from the session rate • Pattern similar to the trace data July 18, 2012 Peer-to-Peer Television for the IP Multimedia Subsystem 91

Workload Evaluation • Session event error : Online event Channel off event Channel on event Less than 1. 5 % of errors are greater than 1 second Offline event Session event error • Approximations introduced by the synthesis algorithm • Very small, do not substantially affect the workload July 18, 2012 Peer-to-Peer Television for the IP Multimedia Subsystem 92

Simulwatch • Session length • Use the Simulwatch model parameters by Qiu et al. [1] Parameter p 1 p 2 p 3 don 0. 3 0. 66 0. 04 doff 0. 19 0. 75 0. 06 dch 0. 23 0. 64 0. 13 λ 1 λ 2 λ 3 1. 3 � 10 - 3. 3 � 10 2. 3 � 102 3 4 3. 2 � 10 - 2. 5 � 10 - 2. 4 � 102 2. 1 3 4 2. 6 � 10 - 3. 2 � 102 3 [1] Qiu et al. , Modeling user activities in a large IPTV system, 2009 July 18, 2012 Peer-to-Peer Television for the IP Multimedia Subsystem 93

Simulwatch • Session length • Use the Simulwatch model parameters by Qiu et al. [1] Parameter p 1 p 2 p 3 don 0. 3 0. 66 0. 04 doff 0. 19 0. 75 0. 06 dch 0. 23 0. 64 0. 13 Online session length λ 1 λ 2 λ 3 1. 3 � 10 - 3. 3 � 10 2. 3 � 102 3 4 3. 2 � 10 - 2. 5 � 10 - 2. 4 � 102 2. 1 3 4 2. 6 � 10 - 3. 2 � 102 3 Offline session length [1] Qiu et al. , Modeling user activities in a large IPTV system, 2009 July 18, 2012 Peer-to-Peer Television for the IP Multimedia Subsystem 94

Simulwatch • Session rate • Power spectrum is a Weibull distribution • Individual spikes at frequencies of 1 hour, 30 min, 15 min k μ Parameter xon July 18, 2012 0. 003 6 278 Peer-to-Peer Television for the IP Multimedia Subsystem 1 hour 30 min 1. 76 1. 41 95

Simulwatch • Session rate • Power spectrum is a Weibull distribution • Individual spikes at frequencies of 1 hour, 30 min, 15 min k μ Parameter xon 0. 003 6 278 1 hour 30 min 1. 76 1. 41 No phase Session rate spectrum July 18, 2012 Session rate Peer-to-Peer Television for the IP Multimedia Subsystem 96

Session Coordination Without going into much details… Keep the channel blocking ratio at a desired reference However… • The blocking ratio is a slow metric • Cannot respond effectively to spikes in the channel activity • Output: 1 Fast control loop 2 Slow control loop • Reference: • Output: • Reference: Desired blocking ratio • We assume the systems are linear and time invariant • July Proportional-integral 18, 2012 Peer-to-Peer controllers Television for the IP Multimedia Subsystem 97

Session Coordination • Priority versus non-priority mode • The slow control loop drives the blocking ratio b(k) toward the reference β • Non-priority: disable the control loop when Allow blocking ratio smaller than the reference Priority mode July 18, 2012 Non-priority mode Peer-to-Peer Television for the IP Multimedia Subsystem 98

Session Coordination • Priority versus non-priority mode • It has a limited impact on session utilization • Non-priority mode preferred for better performance Priority mode July 18, 2012 Non-priority mode Peer-to-Peer Television for the IP Multimedia Subsystem 99

Peer Coordination • Peers are organized in peer pools Global pool 1 All peers with free bandwidth 2 Stream pool 3 Inactive pool Peers with free bandwidth downloading the corresponding stream Peers with inactive sessions for the corresponding stream • Double mapping between peer and its resource 0. 0 0. 1 0. 2 0. 3 0. 4 level Resource 5 1 9 2 5 8 0 5 5 6 8 1 001 055 098 124 354 364 457 612 791 830 874 965 July 18, 2012 Peer-to-Peer Television for the IP Multimedia Subsystem level Peer identifier 100

Peer Coordination • Peer selection based on pool membership Channel connect Channel recovery Inactive session Inactive pool Stream pool Broadcast server Global pool Broadcast server July 18, 2012 Peer-to-Peer Television for the IP Multimedia Subsystem 101

Hybrid Streaming • Create a model of the bandwidth usage • Function of the number of multicast channels set M and unicast channels set U Server Core network multicast Stream Channel bandwidth popularity Unicast path tree Multicast length size Bandwidth Core network unicast Access network Peer-to-peer overhead Access network upload Access network download July 18, 2012 Peer-to-Peer Television for the IP Multimedia Subsystem 102

Hybrid Streaming • Create a model of the bandwidth usage • Function of the number of multicast channels set M and unicast channels set U Primary streams Secondary streams Multicast streams Bandwidth Access network upload Access network download July 18, 2012 Peer-to-Peer Television for the IP Multimedia Subsystem 103