# Calculating the model complexity for wireless sensors

I am coming from geoinformatic background and I do not know anything about the electrical issues.

I have a model for outlier detection for wireless sensor and I want to assess the energy complexity in terms of computation and communication and storage. My model uses the data that were stored before and fit the model and observe every 2 minutes data and aggregate them and send massage to other neighbor point when identify outlier.

I have proposed to use a piece of information plz guide me to interpret these information in order to calculate the complexity.

TICC2420 is programmable in eight levels (from  approximately –25 to 0 dBm)
Mac layer   IEEE 802.15.4
Transmit bit rate   250 kbps
Operation frequency     2.4 GHz
Packet size 128 bytes
Transmit current at 0dBm    17.4 mA
Transmit current at -25dBm  8.5 mA
Supply voltage  (1.6 – 2.0 V)

E=P×T

P=I×V

T=(Packet Size)/(Transmit bit rate)


-

I have some real data for my TH12 temperature and humidity sensor which reports it's data every two mins. (using 802.15.4):

https://github.com/malvira/th-12/wiki/wake-current

https://github.com/malvira/th-12/wiki/th12-sleep-current.ods

I'm not entirely sure what you are calculating but make sure to keep your units straight (it's easy to get turned around).

The numbers you have calculated are Power (so J/s) Ptx and Prx (current * voltage). Keeping track of how long things are happening will be very important.

If you look at my data, there is significant overhead time that has to do with the practical details of the system and so you can't just take the payload length and multiply by the bit rate to get how long the TX is on for.

For instance, my CPU is on for 180mS @ 6mA while it comes out of sleep and the control process gets activated by the time slice in the OS. Then the sensor needs about 1000mS @ 1mA to take the reading.

The TX only takes 7ms @ 24mA but there is 120mS @ 17mA of overhead because of how the radio driver works.

In short, power consumption is quickly dominated by other factors that can't really be estimated without having a benchmark system working.

-