1/7/2022

Sensor node hardware

Augmented general-purpose computers

PC104, Sensoria WINS NG, PDAs

om
Commercial off-the-shelf (COTS) OS & components: Wndows CE,
Linux, Bluetooth, IEEE 802.11
Dedicated embedded sensor nodes

l.c
Berkeley motes, UCLA Medusa, Ember nodes, MIT µAMP
COTS chip sets (small form factors), programming languages(e.g., C)
Sensor Network Platforms and Tools

ai
System-on-chip (SoC)
Smart dust, BWRC picoradio nodes, PASTA nodes

gm
s@
Berkeley Motes –Mica Mote
pd Power Consumption of Mica Motes
am
ya
iv
dh

1
 1/7/2022

Sensor Network Programming Challenges

Traditional programming models not suitable

om
Programmer must handle with messaging, networking, event synch.,
interrupts etc.
Embedded OS (if any) expose hardware details to programmer

l.c
Distributed algorithms/data structures difficult to implement
Respond to multiple stimuli quickly

ai
gm
s@
pd
am
THANK YOU
ya
iv
dh

2
 1/7/2022

Tiny OS

Targeted for resource constrained platforms (motes)

om
Small memory footprint
No filesystem
Only static memory allocation
Software made up of components

l.c
Components create tasks and added to task scheduler
Tasks run to completion: no preempting by other tasks
Software Platforms

ai
Events: interrupts from hardware
Run to completion, can preempt tasks

gm
s@
nesC
pd TinyGALS
am
Extension of C for TinyOS
Dataflow language
Components
Programming model
Interface
Supports all TinyOS components
ya

Defines what functionality component uses and provides

Construct asynch. actors from synch. components
Implementations
Construct application by connecting asynch. components through FIFO
Modules: written in C-like syntax
iv

queues
Configurations: connect interfaces of existing components
Code Generation
Cuncurrency
dh

Map high-level constructs to low-level code for motes

Asynchronous code (AC) vs. Synchronous code (SC)
Automatically generate code for scheduling, event handling, FIFO
SC atomic w.r.t. other SC
queues
Programmers must understand concurrency issues in code

1
 1/7/2022

Node-Level Simulators Node-Level Simulator: ns-2 & TOSSIM

Sensor node model ns-2

om
Mobility of nodes Originally developed for wired networks
Energy consumption Extensions for sensor nodes
Communication model Node locations vs. logical addresses
Capture details of communication (RF propagation delay, MAC layer Energy models

l.c
etc.) Physical phenomena
Physical environmental model TOSSIM

ai
Model physical phenomena in operating environment Simulator for TinyOS apps on Berkeley motes
Statistics and visualization Compiles nesC source into simulator components

gm
Collect results for analysis

s@
State-Centric Programming
pd
am
Applications more than simple distributed programs
Applications depend on state of physical environment
Collaboration Groups THANK YOU
ya

Set of entities that contribute to state updates

Abstracts network topology and communication protocols
iv

Multi-target tracking problem

Global state decoupled into separate pieces
dh

Each piece managed by different principal

State updated by looking at inputs from other principals
Collaboration groups define communication and roles of each principal