1. Modified OSI Architecture for Low-Power Wireless Networks
Jay Bruso
Mike Matranga

2. Agenda Review Thesis Physical Layer Fundamentals
Modulation / Demodulation
Front-end circuitry
PLL
Data Link Layer Fundamentals
Methods of data processing
Proposed solution
Project objectives

3. Thesis Proposal
Applying standard OSI model to WSNs results in poor energy utilization
Physical Layer lacks knowledge of data being sent across the channel
Packet Format, Length, Data Type, etc.
Data Link Layer forced to do work that could be done in Physical Layer
Cost of performing work in Data Link Layer will be more expensive than equivalent function in Physical Layer
For energy optimization
Physical Layer needs to be smarter

4. Transceiver Block Diagram

5. Modulation to the CC2400 Frequency Shift Keying
The CC2400 uses FSK/GFSK

6. Receiver Front End Review Part 1
MIXER
RF AMP
RF IN
RF IN * A
IF OUT
LO IN
RF amplifier: amplifies a weak RF signal coming out of the antenna.
Mixer: together with the local oscillator downconverts the RF signal
to the IF frequency band.

7. Receiver Front End Review Part 2
Power Divider
Frequency Synthesizer
Control
Logic To
Mixers
N, R
from
TXCR
Control
Logic
SPI from Microcontroller 90
CLK to Microcontroller
Reference Clock
Frequency Synthesizer is a PLL

8. Phase Locked Loop N and R are programmed by the microcontroller,
and stored into registers in the transceiver.

9. Receiver Front End Review Part 3
Filter
AGC
ADC
RSSI
CLK
Bandpass Filter: Removes out of band signals
Automatic Gain Control (AGC):
Adjusts the IF amplifier gain according to the signal level
(to keep the average signal amplitude almost constant)
Analog to Digital Converter : Digitizes the analog signal

10. Demodulation V1 V2 PLL Tracking Voltage FSK Signal In 1
This data is digitally filtered & buffered.
Once the signal is recognized, the outputs
CLK, SOM, RXDATA are sent to the Microcontroller

11. Typical WSN Block Diagram
Controller configures XCVR for receive mode
XCVR notifies controller with SOM signal
Controller processes incoming data
Controller disables XCVR when done

12. Methods of Data Processing
Method 1, Polling for Bytes
Controller continuously checks for new data received
Checks if message is complete, and if so, powers off transceiver
Most inefficient because CPU always running

13. Methods of Data Processing
Method 2, Interrupt Driven
Controller interrupted after every byte is received
Checks if message is complete, and if so, powers off transceiver
More inefficient than polling because CPU can go to sleep (enter low-power) mode in between receiving data

14. Methods of Data Processing
Method 3, DMA
Controller interrupted after a block of data is received
More efficient than interrupt because CPU is woken up less often
More difficult than interrupt because size of DMA transfer must be predetermined
Difficult when receiving packets of unknown length
Usually results in setting DMA transfer size small (5 bytes, etc)

15. Data Processing Comparison
Method
Response Time
Power
Consumption
Polling
High
Interrupt
Low
DMA
Med

16. Where is Power Wasted
Finite amount of CPU time required to process and detect end of message
Finite amount of time required for CPU to disable power to XCVR
Wasted Power = (Tproc+Tpwroff)*Ireceive

17. What Can Be Done
Provide physical layer with intelligence for detecting EOM condition
Packet Format
Message length encoding
Offset into data stream where to find length
Character frame encoding
Provide register for escape character (0x80)
Provide register for EOM character (0xFE)
No Data Received Timer
Automatically powers off if no message received within programmed time interval

18. Focus of Project Quantify how much power is really being wasted
Design and implement a small WSN
Measure power wasted
Various protocols
Various processing methods
Various data rates
MSP430 as controller
ChipCon CC2400 as transceiver

19. Focus of Project (cont’d)
Quantify how much power can be saved if physical layer were smarter
Implement proposed solutions in small CPLD (XPLA3 … XCR3512XL)
Repeat previous tests and measure power saved
Various protocols
Various processing methods
Various data rates

20. Project Model Add DLL functionality into PL using CPLD
Demonstrate improved performance
Quantify energy savings
Report results