Spread spectrum trades bandwidth for robustness. Multiply the data by a fast pseudo-random code and it smears across a wide band at low power density; multiply again at the receiver and it collapses back — while interference gets spread away, and a sharp correlation peak reveals exactly where the packet is.
A spread-spectrum transmitter multiplies its narrowband data by a much faster pseudo-random code (a PN sequence). The result occupies far more bandwidth than the data needs, at a correspondingly lower power spectral density — the signal is spread thin across the band, often below the noise floor. The ratio of the spread bandwidth to the data bandwidth is the spreading factor.
This looks wasteful, and in raw spectrum it is. What it buys is robustness: a signal spread wide and low is hard to jam, hard to intercept, tolerant of narrowband interference, and able to share a band with other coded users (the basis of CDMA). The bandwidth is the price; resilience is the product.
The receiver multiplies the incoming signal by the same pseudo-random code, aligned in time, and integrates — it correlates. For the wanted signal the code cancels itself and the data collapses back to narrowband, concentrating all its energy. For anything uncorrelated with the code — noise, other users, a narrowband interferer — the multiplication spreads it out instead, so only a small fraction lands in the data bandwidth.
That asymmetry is the processing gain: the wanted signal is gathered up while interference is dispersed, improving the effective signal-to-interference ratio by the spreading factor. A narrowband jammer that would swamp the raw data becomes a faint, spread-out background after despreading.
Despreading only works when the receiver's code is aligned with the transmitter's. When it is, the correlation produces a sharp peak; when it is not, the output is low and noise-like. So the same correlation that despreads the data is also how the modem acquires: it slides the code in time looking for the peak, and the peak's position gives both the presence of a packet and its exact timing.
This is why spread-spectrum systems are good at detecting a packet at very low SNR — the correlation concentrates energy that is otherwise buried. Preamble detection, timing recovery and interference rejection all fall out of the one operation.
The cost is bandwidth and correlation effort. The receiver must search over code phase (and often frequency) to find the peak, which is a lot of multiply-accumulate at the chip rate — done with matched filters or parallel correlators in hardware. Longer codes give more processing gain and better detection but more to search and more to implement.
For a modem designer the work is the correlator: enough parallelism to acquire quickly across the search space, at the chip rate, inside the power budget — and the code design that balances processing gain, acquisition time and how many users can share the band. Get it right and a weak, spread signal is pulled cleanly out of noise and interference.
A technique that multiplies narrowband data by a fast pseudo-random code to spread it across a much wider bandwidth at low power density. It trades spectrum for robustness: resistance to jamming and interception, tolerance of interference, and the ability to share a band with other coded users.
The improvement in signal-to-interference ratio from despreading, equal to the spreading factor (spread bandwidth divided by data bandwidth). Despreading gathers the wanted signal's energy while spreading interference out, so only a fraction of the interference lands in the data bandwidth.
The receiver multiplies the incoming signal by the same pseudo-random code, time-aligned, and integrates — a correlation. The wanted signal collapses back to narrowband while uncorrelated interference is spread out and suppressed.
Through the correlation peak. Despreading produces a sharp peak only when the receiver's code is aligned in time with the transmitter's, so the modem slides the code looking for the peak; its position reveals both that a packet is present and its exact timing.
Because the despreading correlation concentrates the wanted signal's energy that is otherwise buried in noise, producing a detectable peak even when the signal sits below the noise floor. Interference, being uncorrelated with the code, is spread out instead.
Bandwidth, and correlation effort. The signal needs far more spectrum than the data, and the receiver must search over code phase (and often frequency) to find the peak, which is a large amount of multiply-accumulate at the chip rate implemented in hardware correlators.
Acquisition by correlation.
KNOWLEDGEThe other robustness layer.
KNOWLEDGEWhere LPWA spreading fits.
KNOWLEDGEGPS is spread spectrum.
PRODUCTA correlating modem.
KNOWLEDGEWideband correlation for range.
A correlator with enough parallelism to acquire quickly at the chip rate inside the power budget, and a code that balances processing gain against acquisition time, are the real work. Tell us the waveform.