CI-OFDM
Understanding CI-OFDM
Standard OFDM transmits independent data symbols on N orthogonal subcarriers via an inverse FFT. When many subcarriers happen to align in phase, their amplitudes add constructively, creating a peak N times the average amplitude. The resulting PAPR of 10·log10(N) dB (12 dB for N = 16, 20 dB for N = 100) forces the power amplifier to operate with large back-off from its compression point, wasting DC power. CI-OFDM addresses this by pre-multiplying each subcarrier by a phase code ck,n = ej2πkn/N + jφn before the IFFT. The CI codes are designed so that the resulting time-domain signal has a more uniform envelope, with peaks occurring less frequently and at lower amplitude.
The key advantage of CI spreading over other PAPR reduction methods (clipping, SLM, PTS) is that it requires no additional IFFT computations, no side information, and introduces no signal distortion. The receiver simply applies the conjugate CI code to each subcarrier before equalization, recovering the original data symbols. In frequency-selective channels, CI spreading creates inter-carrier interference that requires an MMSE or SIC receiver for optimal performance, adding modest complexity compared to the simple per-subcarrier equalization of standard OFDM. For dual-function radar-communication (DFRC) systems, CI-OFDM is particularly attractive because the more uniform time-domain waveform also produces better range sidelobe properties than standard OFDM, improving radar performance while maintaining communication data throughput.
CI-OFDM Spreading and PAPR
ck,n = ej2πkn/N for user k, subcarrier n
OFDM PAPR (worst case):
PAPROFDM = 10 · log10(N) [dB]
PA Efficiency Gain:
ηavg ≈ ηmax · 10-backoff/10 [approximate]
Where N = number of subcarriers, k = user/code index, ηmax = PA efficiency at P1dB. CI-OFDM reduces required back-off by 3 to 6 dB, improving ηavg from 12% to 30% for a typical Class AB GaN PA.
PAPR Reduction Method Comparison
| Method | PAPR Reduction | Extra IFFT | Side Info | BER Penalty |
|---|---|---|---|---|
| CI-OFDM | 3 to 6 dB | None | None | None (ideal) |
| Clipping + filtering | 3 to 5 dB | None | None | 0.5 to 1 dB |
| SLM (U = 4) | 2.5 to 3.5 dB | U × IFFT | log2(U) bits | None |
| PTS (V = 4) | 3 to 4 dB | V × IFFT | V phase bits | None |
| DFT-s-OFDM (SC-FDMA) | 4 to 6 dB | Extra DFT | None | None |
Frequently Asked Questions
How does carrier interferometry reduce OFDM PAPR?
CI codes pre-multiply each subcarrier by a pseudo-orthogonal phase rotation, ensuring peaks rarely align simultaneously. This reduces the CCDF probability of high peaks by 3 to 6 dB. Unlike clipping, CI spreading is reversible at the receiver by applying the conjugate code, causing no signal distortion or BER degradation.
What is the benefit for power amplifier design?
OFDM's 8 to 12 dB PAPR forces large PA back-off, reducing efficiency to 10 to 20%. CI-OFDM's 3 to 6 dB reduction allows operation closer to saturation, improving average efficiency from 12% to 30% for a Class AB GaN PA. This cuts DC power by 60% for the same output, critical for 5G massive MIMO with 64 to 256 PA chains.
How does CI-OFDM compare to other PAPR techniques?
Clipping introduces distortion (0.5 to 1 dB BER penalty) and spectral regrowth. SLM/PTS need multiple IFFTs and side information. CI achieves 3 to 6 dB reduction with one IFFT and no side info. The trade-off is that CI spreading requires MMSE or SIC receivers for optimal performance in frequency-selective channels.