The nesC Language: A Holistic Approach to Networked Embedded Systems

1
The nesC LanguageA Holistic Approach to
Networked Embedded Systems
Collect
2
Outline

• Why the need for such systems?
• What are Motes?
• What is the holistic approach
• Overview of NesC
• TinyOS
• NesC Design
• What the future beckons?
• Strengths of the paper
• Weaknesses of the paper

3
Why the Need?

• Getting aggregated data
• Monitoring devices ( earthquakes, natural
  calamities)
• Vehicle detection( information highways)
• Rescue missions( Firefighters)
• Its all about the money
• Commoditization of radio hardware

4
What are motes?

• Small Low powered computers
• Analogous to smart dust
• Reactive
• Data aggregation
• Spatial Localization
• Cheap ( range of 50)

5
What is the holistic approach?

• Guesses?

6
What is the holistic approach?

• Applications deeply tied to hardware
• Resources known statically
• Reusable components used.
• Hardware/Software boundary varies

7
Overview of NesC

• Language for networked embedded
• Systems( used by more than 100 groups)
• Integrates reactivity to the environment
• Concurrency
• Communication Event Driven Systems
• Balance between program analysis and expressive
  power

8
TinyOS

• Component based Architecture
• Tasks and event based concurrency
• Split Phase operations
• Request and Response indicated separately.

9
nesC Design

• Extension of C
• Whole program analysis
• Static Language
• Supports and reflects TinyOS design

10
nesC Design ( More)

• Component Specification
• Component Implementation
• Concurrency and Atomicity
• Concurrency in NesC
• Parameterized Interfaces

11
Comparison of NesC with other languages

• Interfaces
• Modula-2 and Ada No Explicit binding of
  interfaces
• ArchJava2 has bidirectional interfaces
• Modules
• Analogous to Mesa Modules and configurations
• Encapsulation
• Architecture constructed by encapsulation (VHDL)
• Dynamism
• C, BOOPL Registration of event handler with
  event producer.
• Concurrency
• ESC Enforce ordering contraints
• LockLint Focus on locking rather than atomicity

12
Applications

• TinyDB

13
TinyDB

• Entire sensor network as one single,
  infinitely-long logical table sensors
• Columns consist of all the attributes defined in
  the network
• Multiple concurrent queries
• Network Monitoring
• In network persistent storage

SELECT MAX(mag) FROM sensors WHERE mag gt
thresh SAMPLE PERIOD 64ms
App
Query, Trigger
Data
TinyDB
Courtesy http//telegraph.cs.berkeley.edu/tinydb/
tinydb_tutorial_icde.ppt
14
Strengths of NesC

• Inspired by hardware, not tied to it
• Not a heavyweight language as has minimal
  functionality
• Helps reduce memory footprint
• Lack of dynamism helps.
• Static concurrency checking

15
Disadvantages of NesC

• Not suitable for embedded, hard real time control
  systems.
• Not suitable for large scale embedded systems
  e.g. VxWorks, WinCE etc.
• Unsatisfaction with Parameterized interfaces
• Might not be suitable for dynamic motes or
  embedded devices that move between networks.(
  Events)

16
Other Applications

• Timber as an OS for real time embedded systems
• Timed - embedded Reactive
• Real time, reactive, concurrent, object
• oriented language
• Timber objects include methods vs. tasks and
  communication/event handlers.
• Enable the scheduler to intervene execution at
  occurrence of interrupt.
• Dynamic storage allocation
• Direct real time support.

17
What the future beckons?

• Better mechanisms to implement multi client
  services
• Explicit support for FSM- style decompositions
• Alternate implementations of atomic sections

18
2020

• What does this look like?
• Any guesses

"Certainly within the next 10 years we could
wear 20 of garments with electronic components
in them." , Joe Twist, BBC news online
Chips everywhere
Problems Million dollar question?? Privacy
19
The Future is here

• E-watch
• Stress Tester
• Data aggregated and sent to central computer

Smart Cars
Computer Assisted Physical Rehabilitation
20
Whats the fuss all about?

• Body Area Networks
• Low power devices
• Scavenging for power
• Use low power radio (UWB)
• Needs more research to be accurate( cant afford
  to miss a heart attack)

21
Support for Middleware

• Working on this one

22
Strengths of the Paper/NesC

• Author provides substantial evidence of
  performance improvement using features of NesC
• Provides comparison with other real time systems
  and how NesC is better

23
Weaknesses of the Paper

• Not really given any timeline as to what preceded
  NesC and what was the need for such a
  specialized language.
• Could have given real world applications(
  attractive to an end user)

24
Relevance to embedded systems
HOLD YOUR BREATH!
http//video.google.com/videoplay?docid4993846989
748115450total39start0num10so0typesearch
plindex4
25
References

• 1 http//citeseer.ist.psu.edu/cache/papers/cs2/5
  04/httpzSzzSzwww.sics.sezSzrealwsn05zSzpaperszSzk
  ero05timber.pdf/timber-as-an-rtos.pdf
• 2 http//www.jneuroengrehab.com/content/2/1/6
• 3 http//telegraph.cs.berkeley.edu/tinydb/
• 4