Should and Can a Communication System MSOffice1 Adapt Pervasively An Unofficial View http://san.ee.ic.ac.uk Erol Gelenbe www.ee.ic.ac.uk/gelenbe Imperial College London SW7 2BT e.gelenbe@imperial.ac.uk
Slide 1 MSOffice1 ; 26/06/2003
Central Issues Users are highly mobile Internet-Phone Convergence? Needs are context dependent [context of time, place, task, environment..] Terminals are complex and compact Resource opportunities are highly dynamic in space, time, capabilities, prices Systems must be robust and selforganising at all levels surface and deep Systems must be modular and adaptive
Beyond 3G Telephony
Fixed and Mobile Convergence
A Networked System: The Family s Private Network (FPN) It is a User from some perspectives and a Service for the Family - Is private to family members and specific services: Home Security, the Housekeeper, the Parents,.. - Is fully known to itself (self-aware, self-monitoring) - Has to be cheap, secure, private and dependable - Must dynamically and economically use all available communication modalities including sensor networks, IP etc., public fixed and mobile telephony, WiFi,.. You guess it.. - Children and Parents have wireless PDA s - Children use home learning center based on multiple external services - Parents cars are connected - Infants and grand parents are being monitored - Teen-agers location is monitored - The house security system is on-line - Home temperature, sound sensors & actuators are on-line - Home appliances, security, entertainment, heating, lighting are monitored and controlled - Autonomic Self-Aware Management of Connections, Costs, QoS, Policies, Virtual Services and Network Resources, Monitoring, Upgrading..
What Kind of Terminals?
What Kind of Traffic?
Simple Trends in the Internet -The IP Core improves.. and offers increasing performance (e.g. Bandwidth, QoS), but the Internet remains largely unknown to the user - The Periphery of the Internet: the Barbarians and Rome Hot Stuff happens at the periphery of IP, both in a parasitic and symbiotic relationship Profit is generated by services, not at the Core Users and Services cannot directly know network state can they discover it? - Peer(s)-to-Peer(s) Networking on Wireline & Wireless 3G, 802.11, - PDAs, Cellular Phones, Pervasive Networks - Networks of Users, Services, Repositories, Sensors, Actuators, Real or Simulated Robots - Multimedia & Multiple Services - The hope for Seamless & Reliable Communications, QoS, Security, Justifiable Billing
MSOffice15 Internet Architecture in Reality: An Assembly of Inter-dependent Protocols - The Web is a possible Standard User Interface - TCP the Transmission Control Protocol: Controls Packet Flow for a Connection as a Function of Correctly Received or Lost Packets (TCP Reno, Vegas, etc.) & Retransmits Lost Packets - BGP: Determines Paths between Clouds of Routers belonging to Autonomous Systems (AS) - MPLS: Carries out fast Packet Switching based on Predetermined Paths within ASs using Labels, and Implements Traffic Engineering within ASs - IP (Internet Protocol) Implements Shortest Path Routing within ASs. Variants of IP Address QoS (e.g. DiuffServ, IPV6), Weighted Fair Queueing, Congestion Control through Packet Drop
Slide 9 MSOffice15 ; 26/06/2003
MSOffice10 Critique from the Founding Fathers (DARPA) in August 2004 - Flaws in the basic building blocks of networking and computer science are hampering reliability, limiting flexibility and creating security vulnerabilities (Note that DARPA paid for most of these developments!!) - DARPA wants to see the IP and the OSI protocol stack revamped - The packet network paradigm needs to change we must have some mechanism for assigning capabilities to different users today s networks are stationary and have a static infrastructure (mobile) nodes should be able to automatically sign on to networks in their vicinity
Slide 10 MSOffice10 ; 26/06/2003
MSOffice11 DARPA s Main Points - Assigning capabilities to different users How to offer Quality of Service and Service Level Guarantees to Different Users and User Classes, and how to Monitor the Outcome - (Mobile) nodes should be able to automatically sign onto networks in their vicinity - Specific Problems of Sensor Networks, either Wired or Wireless: Networks with Intermittent, Highly Bursty and Urgent Needs
Slide 11 MSOffice11 ; 26/06/2003
DARPA Vision Control Plane Architecture (e.g. Lockheed-Martin) Control Planes IN A NUTSHELL Meta Layer (Claim 6) Meta Optimization Probe Distribution Adaptive Measures Data/Functional Allocation Emergent Behavior Strategic Layer (Claim 5) Network Control Predictive Measures Global Optimality Tactical Layer (Claims 1,2,3,4) Global Performance Enhancement Based on Predicted Traffic Mission Data Host Feedback Path Selection Based on Current Packet Measures Traffic Metrics Shape Optimization Traffic Control Leverage Future Router Technology KEY WORK ALLOCATION: Central Florida University Cognitive Packet Networks San Diego State University Packet Shaping Aristotle University Global Optimization Lockheed Orincon IA CISCO
Internet of the Future: CPN in the Control Plane Architecture (cf. Lockheed-Martin) Military Networks and and Internet ACK Packets Smart Packets Payload Packets Tactical Cognitive Route Control Cognitive (RNN) Route Selection Store/Forward Buffer Upstream Filter Assertion Traffic Engineering Connection Management Multi-homing Path Selection Relay, Route Splitting Strategic Route Modeling and Global Optimization Precedence Based Routing Learned link fidelity models Filter priorities Strategic Link Quality Prediction Mission Awareness Emergent Behavior Analysis Meta
The Alternative: On-Line Measurement and Adaptation Autonomic Networks QoS techniques based on parameter setting such as IntServ, DiffServ and IPv6 have not been successful Users should dynamically formulate their QoS Needs Services should formulate their Capabilities and Costs Both should be monitored and evaluated, The Network is then constituted of groupings of Services and Resources to respond to Users, monitor the outcome and adaptively re-create groupings Network software should be agent based and modular The system should self-organise
Back to Users and Services
User s Needs & Situational Awareness
Autonomic Networks Technical Challenges Feasibility, Efficiency, Cost, Acceptance How can Users, both simple and complex, dynamically formulate their QoS Needs How can services formulate their Capabilities and Costs in a simple unambiguous and universally acceptable manner Is monitoring and evaluation third party based? What is the business model? How can discovery be conducted, and is it possible? How should control information be found and/or disseminated? Security and Reliability Is this all just a collection of P2P overlay networks?
As Network Services Converge to a Common Framework Opportunities for Network Attacks Increase Consequences Get Worse 1996. Panix 1998. Analyzer attacks the Pentagon 2000. Mafiaboy attacks Amazon, Yahoo etc. 2001. Port of Houston 2002. 13 Root Servers 2003. American hackers attack Al Jazeera
Search with Perfect Ignorance Average Travel Time E[T] vs Avg Time-Out 1/r for a regular 8 neighbour grid with d=1 and D=10 (b=0, c=1.5)
Time to Travel to a Destination Which is at Distance D Average Time to Search E[T] Drift b, Second Moment Param. c Avg Time-Out 1/r, then E[T] = -2D/ [ b - (b 2 + 2rc) 1/2 ]
Average Travel Time vs Time-Out Rate r With packet Losses
`âv{ exåt Çá àé ux WÉÇx 44 http://san.ee.ic.ac.uk