Encoding Bits in Waves
Digital modulation maps sequences of bits onto analog waveform parameters — amplitude, phase, or both. The constellation diagram provides the clearest visualization: each symbol occupies a unique point in the I-Q (in-phase / quadrature) plane, and the receiver must determine which point was sent despite the corruption caused by noise, interference, and fading. The art of modulation design is maximizing the number of points (and thus data rate) while keeping them far enough apart to resist errors.
From BPSK to 1024-QAM
Binary Phase-Shift Keying (BPSK) uses just two points — maximum robustness, minimum rate. QPSK doubles to four points at the same error performance by using both I and Q axes independently. Quadrature Amplitude Modulation (QAM) fills a rectangular grid: 16-QAM carries 4 bits per symbol, 64-QAM carries 6, and Wi-Fi 6 pushes to 1024-QAM with 10 bits per symbol. Each doubling of the constellation demands roughly 3 dB more SNR.
Noise and the Error Floor
Additive white Gaussian noise (AWGN) scatters received symbols around their ideal positions. When the noise cloud of one point overlaps a neighboring point's decision boundary, a symbol error occurs. The bit error rate (BER) depends exponentially on the minimum distance between points divided by the noise standard deviation. This simulation adds realistic noise to the constellation so you can see the scatter grow as SNR decreases.
Pulse Shaping and Bandwidth
Real systems cannot transmit infinitely sharp pulses. Raised-cosine filters shape each symbol pulse to limit bandwidth while avoiding inter-symbol interference. The roll-off factor α trades excess bandwidth for time-domain concentration: α = 0 achieves Nyquist-rate bandwidth but requires infinite filter length, while α = 0.35 (common in satellite and cellular systems) adds 35% excess bandwidth for practical implementation.