IQ Sampling & Quadrature Demodulation
To capture a radio signal digitally, we need to record not just how loud it is, but which direction it's pointing — like noting both the length and angle of a clock hand at every moment.
In plain English: IQ data records a signal as two numbers at every instant — like noting both the X and Y position of a spinning dot. Two numbers together tell you both how strong the signal is and which "direction" (phase) it's in, giving you the full picture.
IQ data represents a signal as a pair of values: I (In-phase) and Q (Quadrature). Think of it as describing the position of a point rotating on a circle — instead of saying "the angle is θ", you record its X-coordinate (I = cos θ) and Y-coordinate (Q = sin θ). These two numbers together fully specify both the amplitude and phase of the signal at any instant.
Mathematically, the IQ representation uses complex numbers: the signal is written as I + jQ, where j is the imaginary unit. A pure tone at frequency f₀ appears as a phasor rotating in the complex plane: e^(j2πf₀t) = cos(2πf₀t) + j·sin(2πf₀t). This elegant formulation means every signal processing operation — filtering, mixing, demodulation — can be expressed as complex arithmetic.
In hardware, the I and Q channels are produced by mixing the incoming RF signal with two copies of a local oscillator that are 90° apart in phase. The resulting baseband streams contain all the amplitude and phase information originally encoded at RF, but shifted down to near DC where analogue-to-digital converters can easily sample them. This is the foundation of the superheterodyne and direct-conversion receiver architectures used in virtually every SDR dongle, smartphone, and base station on Earth.
Key Takeaway I and Q together are a complete description of any signal. From them you can recover amplitude (√(I²+Q²)), phase (atan2(Q,I)), and frequency — everything the original signal contained.
In plain English: With a single number stream you can only tell if a signal is above or below your tuning frequency — not which side. With I and Q together, you can tell the difference. That doubles your usable bandwidth and lets you separate signals on either side of your center frequency.
A real-valued ADC sampling at rate f_s can only unambiguously represent frequencies from 0 to f_s/2 (Nyquist theorem). But with IQ (complex) sampling at the same ADC rate, you can represent frequencies from −f_s/2 to +f_s/2 — doubling the usable bandwidth. The "negative frequencies" aren't just a mathematical fiction; they represent signals rotating in the opposite direction on the complex plane, and they correspond to real signals below the local oscillator frequency.
This is why a cheap RTL-SDR dongle with a ~2.4 Msps ADC can receive a 2.4 MHz slice of spectrum centered on any frequency from 24 MHz to 1.7 GHz. The hardware tunes the LO to the desired center frequency, and the complex IQ data stream captures everything within ±1.2 MHz of that center. Increasing the sample rate directly increases the instantaneous bandwidth you can monitor.
Another key advantage: complex baseband signals are easier to process. Bandpass filters become lowpass filters. Frequency shifts become complex multiplications. Demodulation algorithms work on simple arithmetic rather than trigonometric operations. The entire DSP chain — from demodulation to channel coding — benefits from the compact and powerful complex signal representation.
Key Takeaway IQ sampling doubles your bandwidth for the same ADC speed and makes every signal processing step mathematically simpler. That's why every modern radio — phone, WiFi chip, SDR — uses it internally.
s(t) = I(t) + jQ(t)A = √(I² + Q²)φ = atan2(Q, I)BW = f_sampleBW = f_sample / 2s(t)·e^(j2πΔft)