High Performance Communication Korea Embedded Computing MS Program

High Performance Communication
Korea Embedded Computing MS Program
 Textbooks:
There is no textbook required for this class. Electronic copies of research papers and white papers covered in class will be available online. Handouts will be provided for material not available in electronic form.

Select material covered in class will be drawn from the following texts:

• High Performance TCP/IP Networking: Concepts, Issues and Solution, M. Hassan, R. Jain, Prentice Hall, 2003.
• Ad-Hoc Networking Towards Seamless Communication, L. Gavrilovska, R. Prasad, Springer, 2006.

For information on programming the IXP2400s, in addition to the documentation provided with the Software Development Kit, the Intel Press titles on IXP2400/2800 development and programmingare recommended.

Click here to download the files:-

Part I: Networking
The first part of the course will cover networking topics, with a strong emphasis on wireless/ad-hoc networking.

Wireless Networking Overview: May 16, May 18

- Overview of standards, physical and MAC layer, routing protocols. (handout)
- Understanding Congestion in IEEE 802.11b Wireless Networks (reference only)

Cross-layer Optimizations and QoS: May 18, May 23

- Overview of approaches. (Cross-Layer handout, QoS handout)
- An Overlay MAC Layer for 802.11 Networks [slides]
- Supporting Service Differentiation for Real-Time and Best Effort Traffic in Stateless Wireless Ad Hoc Networks (SWAN) (reference only)

Mobility: May 25

- Overview of approaches. (handout)
- Comparison of IP Micro-mobility Protocols
- Towards Realistic Mobility Models for Mobile Ad Hoc Networks (reference only)

Transport: May 30, Jun 1

- Overview of TCP variations. TCP performance in mobile and wireless
- Analysis of TCP performance over Mobile Ad Hoc Networks
- Improving TCP Performance over Mobile Ad Hoc Networks By Exploting Cross-Layer Information Awareness (refernce only)
- Improving TCP Performance over Wireless Networks with Collaborative Multi-homed Mobile Hosts (reference only)
- SCTP. SCTP in wireless. (if interested, here is a tutorial on SCTP)
- SCTP - A Multi-Link End-to-End Protocol for IP-based Networks
- Evaluation of SCTP over Wireless/Satellite Links (reference only)
- A Study of SCTP Services in a Mobile IP Network (reference only)

Part II: Systems and Platforms
The second part of the course will cover topics related to systems software and platform features targeting both end-user embedded devices as well as infrastructure nodes and specialized embedded accelerators (e.g., FPGA, network processors) in such nodes.

OS and Protocol Stack Implementation: June 13, June 15

- OS-level optimization techniques for high performance IO (handout)
- Full TCP/IP for 8-bit Architectures
- Protocol stack splitting for multicore platforms

Power and enegergy efficiency: Jun 22

- Wireless Wakeups Revisited: Energy Management for VoIP over Wi-Fi Smartphones
- Ghosts in the Machine: Interfaces for Better Power Mangement

Virtualization: Jun 27, Jun 29

- High Performance and Scalable IO Virtualization in Self-Virtualized Devices (Sec. 1 and 2)
- A Hypervisor for Embedded Computing
- VirtualLogix approach (reference only)

Programmable Networking: Jul 11

- Click: Modular Router
- A Survey of Programmable Networks (reference only)

Infrastructure Acceleration: Jul 13, Jul 18

- Network Processors, NP overview, applications (handout)
- A Highly Flexible, Distributed Multiprocessor Architecture for Network Processing
- Reconfigurable Hardware, FPGAs, overview, applications (security, pattern matching) (handout)
- SIFT: Snort Intrusion Filter for TCP

Part III: Services, Tools, Applications
The third part of the course will include discussions on various higher-level services (middleware, control-plane services...), benchmarking tools and models, domain specific languages, and classes of applications.

Potpourri: Jul 25

- Efficient and Transparent Dybamic Content Updates for Mobile Clients (except 3.4, 4.1)
- Rethinking Wireless for the Developing World (presentation)
- WiLDNet: Design and Implementation of High Performance WiFi Based Long Distance Networks (reference only)
- A Comparative Study of Data Dissemination Models for VANET (except 3.3)
- CarTel: A Distributed Mobile Sensor Computing System (reference only)
- CogNet - An Architecture for Experimental Cognitive Radio Networks within the Future Internet (reference only)

Services:

- Middleware solutions for wireless, mobile environments. Case study.
- Location-based services.
- Application acceleration

Tools:

- Evaluation: Jigsaw: Solving the Puzzle of Enterprise 802.11 Analysis
- Evaluation and simulation
- Domain specific languages

Specialized Application Domains:

- GPRSWeb: Optimizing the Web for GPRS Links
- Communication in industrial embedded systems
- Communication in vehicular applications
- Sensor networks

 

High Performance Communication

High Performance Communication

Topic/Slides	Assigned Paper/Notes
Class Overview & Introduction
no class (HotI 06)
IXP NPs: Hardware, Software, Programming Model	M. Venkatachalam, P. Chandra, R. Yavatkar, A Highly Flexible, Distributed Multiprocessor Architecture for Network Processing, Computer Networks, 2003 Intel Architecture presentation
IXP NPs: Hardware, Software, Programming Model	Intel SDK presentation Homework 1 out
no class -- Labor Day
1/10 Gigabit Ethernet, Overview and Performance	Feng et al., "Optimizing 10-Gigabit Ethernet for Networks of Workstations, Clusters and Grids: A Case Study" 10 Gigabit Ethernet Expands Network Bandwidth and Shrinks Latency
Infiniband: Hardware	Introduction to InfiniBand Homework 1 due
Infiniband software stack, OpenFabric	TBP Homework 2 out
Proprietary Interconnects: Myrinet, Quadrics, etc. "War of the Interconnects"	Petrini et al., Scalable Collective Communication on the ASCI Q Machine Boden et al., Myrinet: A Gigabit-per-second Local Area Network J. Liu, et al, Performance Comparison of MPI Implementations over InfiniBand, Myrinet and Quadrics
no class	Homework 2 due 9/22
PCI Express, Advanced Switching, Chip-level interconnects	Stargen, PCI Express and Advanced Switching: Evolutionary Path to Building Next Generation Interconnects Project Proposal Due
RDMA, R-NICs ToEs	Allyn Romanow, Stephen Bailey, RDMA over IP Hyong-youb Kim and Scott Rixner, TCP Offload through Connection Handoff, EuroSys 06
NIC design issues Multi-CPU NICs Protocol Offload/Onload discussion	K. Mackenzie, An Intel IXP1200-based Network Interface D.K. Panda, Can User Level Protocols Take Advantage of Multi-CPU NICs?
high performance TCP, SCTP	W. Feng. Is TCP an Adequate Protocol for High-Performance Computing Needs?, SCTP
Communication architectures: Portals	Brightwell, Initial Performance Evaluation of the Cray SeaStar Interconnect (hoti05), (micro06) Portals 3.0
no class (IXA Summit 06)	Obeus.com - The Democratic Ads
no class - Fall Recess
Structuring application communication: MPI, PVM ECho stack (Arvind K.)	Introduction to MPI ECho & EVpath Project Update Report Due
Protocol stack splitting for multi-core platforms (N. Rao B.)	Brecht et al., Evaluating Network Processing Efficiency with Processor Partitioning and Asynchronous I/O
NIC-based acceleration of VMMs (Adit R., Purav S.)	Himanshu Raj, Ivan Ganev, Karsten Schwan, Jimi Xenidis, Scalable I/O Virtualization via Self-Virtualizing Devices Liu et al, High Performance VMM-Bypass I/O in Virtual Machines, USENIX Annual Technical Conference 2006
Programmable networking, Click (Priyanka T.)	Campbell A.T., De Meer H.G., Kounavis M.E., Miki K., Vicente J., and Villela D., A Survey of Programmable Networks Robert Morris, Eddie Kohler, John Jannotti and M. Frans Kaashoek, The Click Modular Router
Network Processor Survey	from Douglas Comer's book
Network Processor Survey	from Douglas Comer's book
Caching in NPs NP Design Analysis (Ranjith S.)	J. Mudigonda, H. M. Vin and R. Yavatkar, A Case for Data Caching in Network Processors Ramaswamy et al., Application Analysis and Resource Mapping for Heterogeneous Network Processor Architectures, NP3, 2004 T. Wolf, Performance Models for Network Processor Design, TPDS
IBM Cell (Vishakha G.)	Cell Moves into the Limelight Cell Multiprocessor Communication Network: Built for Speed
Reconfigurable Hardware, FPGAs (Jeff Y.) Applications: IDS on FPGAs (Gregory D.) Obeus.com - The Democratic Ads	David E. Taylor, Jonathan S. Turner, John W. Lockwood, Dynamic Hardware Plugins (DHP): Exploiting Reconfigurable Hardware for High-Performance Programmable Routers Chris Clark, Scalable Pattern Matching for High Speed Networks
Applications: Classification, Scheduling, Multicast, DHTs, Content Processing	Francis Chang, Wu-chang Feng, Wu-chi Feng, Kang Li, Efficient Packet Classification with Digest Caches, NP3 Laxmi Bhuyan's group research, An Efficient Packet Scheduling Algorithm in Network Processors Design and Implementation of a Content Aware Switch using a Network Processor
no class - Thanksgiving
Tools: Benchmarks (Matt K.)	Byeong Kil Lee, Lizy Kurian John, NpBench: A Benchmark Suite for Control plane and Data plane Applications for Network Processors, Int't Conf. on Computer Design, 2003 NetBench, CommBench
Tools: simulators and development Compiler support (Fernando A.)	Y. Luo, L. Bhuyan, NePSim: A Network Processor Simulator with Power Evaluation Framework, IEEE Micro, Special Issue on Network Processors, September/October 2004 Herbert Bos, et al., FFPF: Fairly Fast Packet Filters Harrick M. Vin, Jayaram Mudigonda, Jamie Jason, Erik J. Johnson, Roy Ju, Aaron Kunze, and Ruiqi Lian, Shangri-La: A Programming Environment for Packet-processing Systems: Design Considerations
Project Presentations	Matt, Jeff, Arvind, Ranjith, Adit
Project Presentations	Priyanka, Vishakha, Gregory, Purav, Rao, Fernando