Lecture 2

Pre-read for Lecture 2
Motivation: Why Probability & Detection Theory Matter in IoT

Pre-read for Lecture 2

Motivation: Why Probability & Detection Theory Matter in IoT

1. The IoT Challenge

Billions of devices need to communicate — many are tiny sensors powered by batteries or energy harvesting.
Devices often transmit short bursts of data (temperature, motion, location).
Signals are weak and easily lost in noise or interference.
Energy efficiency is critical: devices must avoid wasting power while staying connected.

2. Where Detection Theory Fits

IoT devices constantly face decision problems:

Is there a signal or just noise?
Is the channel free or busy?
Is this data normal or an anomaly?

We model these as binary hypothesis tests:

H₀: nothing of interest (noise, idle channel, normal data)
H₁: something important (signal, busy channel, anomaly)

3. Examples in IoT

Wake-up Radios

A sensor node sleeps to save power.
It wakes only if a weak “wake-up signal” is detected.
Wrong decision → wasted energy (false alarm) or missed data (miss).

Spectrum Sensing

Devices share spectrum (e.g., LoRa, NB-IoT).
They must detect if a frequency band is free before transmitting.
Wrong decision → interference or underutilized spectrum.

Anomaly Detection

Smart homes, factories, and cities rely on sensor networks.
Detect if a sensor is malfunctioning or under attack.
Wrong decision → false alarms trigger unnecessary action, or real problems go unnoticed.

4. Why Probability is Essential

Noise and interference are random.
Device decisions can’t be deterministic — they must be based on probability.
Understanding false alarm vs. detection trade-offs is crucial for reliable IoT systems.

Takeaway:
Probability and detection theory aren’t abstract math — they’re the foundation of how IoT devices save energy, avoid interference, and stay reliable.