Towards Ubiquitous Computing

1
Towards Ubiquitous Computing

• An overview of background research academic
  work in networking operating systems design

Doina Bucur doina, Ada-219, 5613 was Computer
Science Department, Politehnica University of
Bucharest, Romania
2
Motivation

• a background in
• networking
• operating systems
• some formal aspects

present
further research
draft PhD proposal Context-Awareness in Global
Ubiquitous Computing
3
The Ubiquitous Scenario...

• Is based on
• heterogeneous computing or sensor hardware
• various embedded operating systems
• tools or protocols for
• communication
• adaptability (and context-awareness)
• performance analysis and protocol design

4
Networking operating systems - the ORC project

• ORC Optimized Route Cache on a Linux router
• Written into the operating system kernel
• Designs and implements a performance-adding
  information routing subsystem, allowing for
  faster data transport
• It mixes issues on operating systems, data
  structures and algorithmic complexity

5
Networking operating systems - the ORC project
(cont)

• Main motivations
• Generic devices including sensors, networking and
  household equipment can use stripped-down Linuxes
  as operating systems. This alleviates the need to
  write a new OS for each platform.
• Modules implementing new features are easy to
  add/remove.
• The software capabilities are no longer dependent
  on the platform.

6
Networking operating systems - the ORC project
(cont)

• A router design requirements
• Eliminate delay in
• packet switching
• path determination
• reachability, next-hop
• MAC header info
• Minimum memory consumption (routing is
  data-intensive)
• If the data flow prefers certain
  source-destination pairs, a cache keeps hold of
  lookup data, to avoid re-looking it up.

7
Networking operating systems - the ORC project
(cont)

• Path determination the process switching method
• Simplest and slowest switching method
• There is a dedicated process for switching
• The forwarding info for every packet is extracted
  from the routing table and ARP table
• The switching process running is driven by the
  scheduler
• Fast-switching
• If packets have preferred destinations, cache the
  info. First packet is process-switched
• The lookup and switching are performed in the
  interrupt handler (no scheduling)

8
Networking operating systems - the ORC project
(cont)

• the output info is part of the tree
• design problem the destination routes kept
  inside the radix tree must be consistent with the
  original info
• cache entries are invalidated if
• original tables change or
• entries age
• Output info no longer memorized inside the table
  (no invalidation)
• The cache is initialized with all info necesarry
  (no packet is process switched)

9
Networking operating systems - the ORC project
(end)

• Cache 256-way tree, prebuilt out of the routing
  table

10
Large-scale network management - the NetWreck
project

• Management Monitoring Control
• Modern networks
• complex systems of various hardware atoms
• different control interfaces
• NetWreck
• topology discovery
• autonomic control (adaptability)
• device control

11
Large-scale network management - the NetWreck
project (cont)

• Autonomic control
• an IBM-coined concept
• conventional management needs too many resources
  (know-how, administrators, time)
• evolution independent self-control, adaptation
• service monitoring and correcting
• security updates
• point to congested areas and reroute traffic

12
Academic areas

• Functional Programming
• lambda calculus
• functional languages
• Formal Languages and Automata
• Algorithms and Algorithm Analysis

13
Academic areas (cont)

• Operating Systems
• processes and process communication,
  synchronization, I/O
• Operating Systems Design
• device drivers
• Security on wireless LANs (whitepapers)
• 802.11 Wireless Networks, Local Area Networks
  with Cisco Systems

14
CS PUB, CalTech, CERN - MonALISA

• MonALISA Monitoring Agents using a Large
  Integrated Services Architecture
  http//monalisa.carc.caltech.edu
• The MonALISA framework provides a distributed
  monitoring service system for Grids. Each
  MonALISA server acts as a dynamic service system
  and provides the functionality to be discovered
  in that context and used by any other services or
  clients.
• The goal is to provide the monitoring information
  from large and distributed systems to a set of
  loosely coupled "higher level services". The
  framework can integrate existing monitoring tools
  and procedures to collect parameters describing
  computational nodes, applications and network
  performance.

15
CS PUB IEEE Computer Society International
Design Competition (CSIDC)

• 2002 first prize BE Secure
• Building Equipment, a sensor surveilling network
  Bluetooth communication
• 2003 second prize Eyes Only
• AES encryption for all data in a PC
• 2004 second prize eXpress! Help
• localization capabilities (GPS) and low-range
  communication (Bluetooth)
• 2005 third prize NOMAD Positioning System
• Bluetooth sensor positioning network

16
Conclusion

• The reasonable man adapts himself to the world
  the unreasonable man persists in trying to adapt
  the world to himself. Therefore all progress
  depends on the unreasonable man. (George Bernard
  Shaw)