Reciprocity-Based DPD

In Time-Division Duplex (TDD) systems, the uplink and downlink share the same frequency band, making the physical propagation channel symmetric in both directions. Reciprocity-based DPD exploits this property by estimating the downlink nonlinear distortion from uplink pilot measurements. The base station captures the composite uplink signal including PA nonlinearities reflected through the reciprocal channel, then derives predistortion coefficients valid for the subsequent downlink transmission slot within the channel coherence time.

Unlike conventional DPD that requires a dedicated feedback receiver physically coupled to each PA output, reciprocity-based DPD derives linearization parameters from far-field radiated signals. The user equipment transmits known pilots; the base station receives them through its own nonlinear receiver chain. By applying channel reciprocity assumptions, the system mathematically separates the propagation channel from the PA distortion to compute the inverse predistortion function for the transmit path.

Traditional massive MIMO DPD requires one feedback receiver per antenna branch, creating prohibitive cost and routing complexity for 64+ element arrays. Reciprocity-based DPD eliminates dedicated downlink observation receivers by reusing the existing uplink receiver chain for distortion estimation. Key benefits:

• Zero additional RF hardware for DPD feedback
• Reduced board area and power consumption
• Simplified calibration and manufacturing test
• Scalable to very large arrays without linear cost growth

The validity of reciprocity-derived coefficients depends on the channel coherence time — the interval during which the physical channel remains essentially static. In high-mobility scenarios, the channel decorrelates rapidly, requiring frequent recalibration. The DPD update rate must be synchronized with the TDD frame structure, typically operating within the guard period between uplink and downlink slots. At sub-6 GHz with pedestrian speeds, coherence times of 5-20 ms provide sufficient margin for coefficient computation.

Channel reciprocity applies only to the physical propagation medium. Transceiver hardware impairments are inherently non-reciprocal — the transmit and receive paths have different amplifiers, filters, and mixers. Reciprocity-based DPD must explicitly model and compensate for:

• Transmit/receive gain mismatch between paths
• Phase noise differences in separate local oscillators
• I/Q imbalance unique to each direction
• Antenna mutual coupling affecting Tx and Rx differently Calibration preambles or factory characterization isolate these non-reciprocal components.

In beamforming arrays, the effective PA nonlinearity varies with the beamforming weight vector due to load modulation and mutual coupling. Reciprocity-based DPD must account for this by estimating distortion under the same beam pattern used for downlink transmission. The uplink pilots are processed with the conjugate beamforming weights to reconstruct the equivalent downlink distortion profile. This creates a beam-dependent DPD coefficient set that must be updated whenever the beam direction changes significantly.