Team• Principal investigator and team leader:
Ahmed Elmagarmid
• Task leaders– Park– Spafford– Ghafoor– Korb– Research team: 18 faculty members
Outline
8:45 Welcome (Sameh)
9:00 Overview (ake)
9:30 Databases(ghafoor)
10:00 Storage(sunil)
10:15 Compression(delp)
10:30 Break
10:45 Networks(park)
11:15 Scheduling(yau)
11:30 Security(spafford)
12:00 Lunch
12:30 Education and Outreach(ake)
1:00 Aplications(enh)
1:15 Infrastructure(korb)
1:45 New faculty(fahmy and Arif)
2:15 Dean visit
3:00 Tour of facilities
Objectives
• Research, education, and outreach in the field of Multimedia Computing.
Networks
Significance
• A unique private high speed backbone that will allow unlimited experimental research to go on with direct access to Purdue rich infrastructure and I2.
Research methods
• Experimental infrastructure for research in:– networks– databases– security– compression– &– storage technology.
Focus
• The use of relevant research in Multimedia with an emphasis on QoS requirements across the various research areas.
MSI LABORATORY CENTRAL FACILITY FOR MULTIMEDIA SUPPORT
INFRASTRUCTURE
CSSystems Software and
Architecture Lab
CSNetwork Systems
Lab
CSRAID Lab
CSMultimedia
Instruction Lab
CSPurdue-on-line
CSCoast Lab
ECEDistributed Multimedia
Software Lab
ECEVideo and Imaging
Processing Lab
Nuclear EngineeringThermal Hydraulic
AndReactor Safety Lab
Veterinary MedicineBasic MedicalScience Lab
Multimedia Research Testbed Construction
• Acquire the equipment necessary to create a world class multimedia computing environment to be used as a multimedia support infrastructure (MSI). The architecture and configuration of this testbed MSI will form the substrate for our research team.
(1) Capture
(2) Compress
(3) Store, Index, Retrieve
(4) Network
(5) Display
Figure 2. Quality of Service Multimedia Infrastructure
Distributed Multimedia Database Management
• Quality of presentation management, specification, and translation as they relate to: data storage, I/O management, data placement, meta-schema design, data replication for reliability and efficiency, admission control for user sessions, and benchmarking of distributed applications.
• Develop and benchmark multimedia compression techniques for the transport of complex multimedia objects across the distributed networks
QoS Management for Networked Multimedia
• We propose to use the MSI infrastructure to perform research in QOS-sensitive access, dissemination, and transport of multimedia data retrieved from distributed multimedia databases.
Security at User and Network Levels
• Security is major concern for information systems that may be deployed in practice. We are planning to use the MSI infrastructure to perform research in ATM security, audit trails, watermarking, and intrusion detection of multimedia networks and data.
Security
• QoS- sensitive security architecture– Security mechanisms (user plane protection,
control/management plane protection).– Secure QoS maintenance (use plane services,
control/management plane services.– ATM security.
Multimedia Capture and Presentation
• Realistic capture, compression, and presentation of multimedia data. To this end, we plan to employ real-world multimedia data applications already accessible to us. The research in this category will be limited to compression. The rest of the issues in capture and presentation are carried out using commercial off-the-shelf technology.
Five Year Summary Budget (in units of $1000)
Purdue Industry NSF Total
Personnel $615 $145 $0 $760
General Expenses $0 $0 $217 $217
Equipment $0 $0 $792 $792
Total $615 $145 $1,009 $1,769
Equipment Category Five-Year BudgetCapture $ 20,400Compression $ 85,000Storage $ 406,000Networking $ 402,500Presentation and Development $ 95,500Total $1,009,400
Table 1: Summary of Equipment by Category
Qty Description Unit Cost Total Cost
3 ATM backbone switch $30,000 $ 90,0008 ATM workgroup switch $ 7,000 $ 56,0004 Gigabit Ethernet switch $12,000 $ 48,0002 Myrinet switch $ 5,000 $ 10,0004 Wireless network $ 1,500 $ 6,0006 WFQ and RSVP router $25,000 $150,00040 Special purpose network $ 1,000 $ 40,000
card1 Multicast streaming $ 2,500 $ 2,500
softwareTotal $402,500
Table 2: Networking Equipment
Qty Description Unit Cost Total Cost
5 MPEG–1 encoder $ 5,000 $25,0002 MPEG–2 encoder $23,000 $46,0002 Digital video to MPEG $ 5,000 $10,000
encoder2 Internet video encoder $ 2,000 $ 4,000Total $85,000Table 3: Compression Equipment
Qty Description Unit Cost Total Cost
2 Hierarchical storage $110,000 $220,000management system
2 Database computer engine $ 55,000 $110,0001 RAID storage server $ 36,000 $ 36,0001 Experimental storage $ 40,000 $ 40,000
serverTotal $406,000
Table 4: Storage Systems and Database Equipment
Qty Description Unit Cost Total Cost
2 Analog video camera $ 1,000 $ 2,0002 Digital video camera $ 3,200 $ 6,4002 Audio recorder $ 1,000 $ 2,0002 Still image recorder $ 5,000 $ 10,00010 Hardware MPEG-2 $ 500 $ 5,000
decode50 Software MPEG-1 $ 50 $ 2,500
decoder50 Internet video decoder $ 20 $ 1,0004 Low-end workstation $ 4,000 $ 16,0004 High-end workstation $14,000 $ 56,0005 Portable computer $ 3,000 $ 15,000Total $115,900Table 5: Capture and Presentation Equipment
Figure 1: Applications, Challenges, and Infrastructure of the Project
QoS Dimensions
• Maximal end-to-end latency levels.• Service precedence among different data sets.• Reliability (e.g. error correction).• Delay (mean transfer delay, 95 percentile transfer
delay). • Throughput (maximum bit rate, mean bit rate).
• Scheduled repeated transmission (number or repetitions, interval of repetitions).
Why is it Important
• Data delivery delay variations, bit errors and data loss all are very pertinent to the quality of the video stream received. A transmission link with a bit error rate of 10-5 would be acceptable for non-real time data transmission with some form of error correction, but unacceptable in a video stream. Therefore, in order to study issues which are significant in terms of the quality of video received we must extend it to QoP.
User Level Quality: QoP
• Oriented toward the perceived quality received and/or specified by the end user for audio, video or multimedia applications– presentation latency,
– jitter,
– luminance levels,
– noise levels in the audio or images,
– truthfulness of color rendering,
– contrast, display resolution in terms of detail/sharpness etc.
Research Agenda
• Networks
• Database
• Security
Networks
• Dual QoS Network Architecture
• Guaranteed Service Architecture
• Stratified Best Effort Architecture
• Experiments
Operating Systems
• Scheduling
• Network/OS interface
MM Databases
• Distributed Multimedia Databases
• Storage technology
• Compression
• Experimentation
Database Issues
• Intelligent Multimedia Data Placement.
• High performance I/O and disk scheduling.
• Tertiary storage issues: FT, Scheduling etc.
Distribution Issues
• Mapping of QoS requirements into terms the DBMS can deal with (admission control, resource allocation etc.).
• QoS-QoP negotiation protocols.
• Video Database servers research and comparison with existing commercial servers.