Why does higher-order modulation require more Eb/N0 for the same BER?

Higher-order modulations pack more constellation points into the same signal space. QPSK has 4 points with wide spacing; 256QAM has 256 points packed much closer together. The closer spacing means less noise is needed to push a received symbol across a decision boundary into the wrong region, causing a bit error. For a target BER of 10^-6 in AWGN: BPSK and QPSK need about 10.5 dB Eb/N0, 16QAM needs about 14.5 dB, 64QAM needs about 18.5 dB, and 256QAM needs about 23 dB. Each step doubles the data rate per unit bandwidth but costs 4 to 5 dB more Eb/N0.

How does FEC coding gain reduce the required Eb/N0?

Forward error correction adds redundant bits that allow the receiver to detect and correct errors without retransmission. A rate-1/2 convolutional code doubles the transmitted bits but provides about 5 dB of coding gain: the same BER is achieved with 5 dB less Eb/N0. Modern turbo codes and LDPC codes approach the Shannon limit, providing 8 to 10 dB of coding gain at BER = 10^-6 with rate 1/2 coding. 5G NR uses LDPC for data channels and polar codes for control channels, achieving within 0.5 to 1 dB of the theoretical Shannon capacity limit. The trade-off is spectral efficiency: a rate-1/2 code halves the effective data rate per Hz of bandwidth.

What is the relationship between EVM and BER?

EVM measures the RMS distance between received and ideal constellation points, while BER measures how often those errors actually cause bit mistakes. For AWGN-dominated impairments, the relationship is approximately: BER = (1 - 1/sqrt(M)) / log2(sqrt(M)) times erfc(sqrt(3/(2(M-1))) / EVM_rms), where M is the constellation order. For 64QAM at 8% EVM (the 3GPP limit), the corresponding BER is approximately 10^-3 before FEC. After LDPC decoding, this can be reduced to below 10^-6. This is why 3GPP specifications define EVM limits rather than BER limits for the physical layer: EVM is easier to measure on the transmitter side without decoding.

Signal Processing

BER

Bit Error Rate

Transmit one million bits over a wireless link and count how many arrive wrong. If 100 are wrong, the BER is 10⁻⁴. That single number tells you whether the link works. Voice can tolerate 10⁻³ (one error per thousand bits). Video requires 10⁻⁶. Financial data demands 10⁻¹². The BER depends on three things: the signal-to-noise ratio at the receiver, the modulation scheme (higher-order modulations pack bits closer together, requiring higher SNR), and the forward error correction code (which trades bandwidth for error resilience). The BER vs. Eb/N0 curve is the Rosetta Stone of digital communications.

Category: Signal Processing

Definition: Errored bits / Total bits

Typical Target: 10⁻⁶ (after FEC)

The Cost of More Bits Per Symbol

Modulation	Bits/Symbol	Eb/N0 for BER 10⁻³	Eb/N0 for BER 10⁻⁶	Spectral Eff.	Application
BPSK	1	6.8 dB	10.5 dB	1 bps/Hz	Satellite, deep space
QPSK	2	6.8 dB	10.5 dB	2 bps/Hz	Cellular (edge-of-cell)
16QAM	4	10.5 dB	14.5 dB	4 bps/Hz	Cellular (moderate SNR)
64QAM	6	14.5 dB	18.5 dB	6 bps/Hz	WLAN, cellular (good SNR)
256QAM	8	18.5 dB	23 dB	8 bps/Hz	Cable, cellular (near BS)
1024QAM	10	22.5 dB	27 dB	10 bps/Hz	WiFi 6E (short range)

BER for BPSK/QPSK in AWGN:
BER = Q(√(2·E_b/N₀)) = (1/2)·erfc(√(E_b/N₀))

FEC coding gain examples:
Rate-1/2 convolutional (K=7): ~5 dB gain at BER 10⁻⁶
Rate-1/2 turbo code: ~8 dB gain at BER 10⁻⁶
Rate-1/2 LDPC (5G NR): ~9 dB gain, within 1 dB of Shannon limit

5G NR adaptive modulation: QPSK at cell edge (Eb/N0 ≈ 2 dB with LDPC) scaling to 256QAM near the base station (Eb/N0 ≈ 15 dB with LDPC).

Common Questions

Frequently Asked Questions

Why does higher-order modulation need more Eb/N0?

More constellation points means closer spacing. Less noise pushes a symbol into the wrong decision region. QPSK needs 10.5 dB for 10⁻⁶ BER. 256QAM needs 23 dB. Each modulation step doubles data rate but costs 4 to 5 dB.

How does FEC help?

Redundant bits enable error correction without retransmission. Rate-1/2 LDPC (5G NR) provides ~9 dB coding gain, meaning 9 dB less Eb/N0 needed for the same BER. Trade-off: halves effective data rate per Hz.

EVM vs. BER relationship?

EVM measures constellation point spread. BER counts actual errors. For 64QAM at 8% EVM (3GPP limit): pre-FEC BER ≈ 10⁻³. After LDPC: <10⁻⁶. 3GPP specs EVM instead of BER because it is measurable on the transmitter without decoding.

Related Terms

← Beamwidth Bias Tee →

Link Analysis

BER vs. Eb/N0 Curve Generator

Select modulation scheme and FEC code rate to plot the BER curve. Compare uncoded and coded performance, and find the required Eb/N0 for any target BER.

Plot BER Curves