NES Research in Sensor Networks The Systems Perspective

1
NES Research in Sensor NetworksThe Systems
Perspective

• Adam Dunkels, PhD

2
The Systems Perspective

• Effects of low-level details on high-level
  behavior
• Building systems with attention to low-level
  mechanisms

3
Activities
int a_protothread(struct pt pt)
PT_BEGIN(pt) PT_WAIT_UNTIL(pt,
condition1) if(something)
PT_WAIT_UNTIL(pt, condition2)
PT_END(pt)

• Energy profiling
• MAC protocols
• Energy, frequencies
• Software development
• Routing
• 6lowpan
• Simulation
• Debugging
• Testing

4
How do we measure energy and energy efficiency?

• Typical method micro-measurements
• Oscilloscope, one node, measure a few seconds
• Predict / extrapolate
• Energy-estimation hardware (SICS, 2005)
• Energy-measurement hardware (Berkeley, April
  2007)
• Software-based energy estimation (SICS, June
  2007)

5
Energy-measurement hardware

• Add-on board (Berkeley, April 2007)
• Costly for network-scale deployment

Jiang et al., IPSN 2007
6
Contiki software-based energy profiling (SICS,
June 2007)

• Zero-cost hardware
• Zero-effort deployment
• Enables network-scale energy profiling
• Good accuracy
• Low overhead

7
Demo see energy in practice
8
MAC protocols

• Reduce power
• Frequency hopping
• Decide who gets to send
• Current activities
• MAC software architecture for Contiki
• Power-saving MAC protocol for ZigBee
• Frequency hopping, decentralized scheduled MAC
  protocol

9
Sensor network programming today
Network/transport/application/
Deluge
VRR
Flush
CTP
Trickle
Application 1
AODV
RBP
GRE
S4
ETX/le
Application 2
ZigBee
Z-MAC
Crankshaft
X-MAC
MAC
B-MAC
S-MAC
SCP
6lowpan
802.15.4
Link
TR1001
nRFxx
CC2420
CC1100
10
The Contiki way
Application 2
Application 1
Rime
Best-effort multi-hop unicast
Reliable bulk download
Best-effort network dissemination
Reliable bulk dissemination
Reliable flooding
Reliable data collection
Chameleon
802.15.4
X-MAC
6lowpan IPv6
Wireless HART
11
Remaining Challenges

• Holistic system optimization
• Separation of system mechanisms from system
  configuration
• Automatic performance tuning
• Performance monitoring
• Robustness and reprogramming
• Binary compression
• Multi-level watchdog

12
Simulation MSPsim
13
Upcoming software

• Contiki 2.1 (this week)
• Hard real-time support
• Rime communication stack
• Power-saving MAC protocols
• Energy profiling
• Contiki 2.2 ( February 2008)
• Chameleon
• Time table-based execution time profiling

14
Contiki

• http//www.sics.se/contiki/