SHAPER (Solutions for wideband Highly lineAr and efficient PowER amplification) is a Software Defined HW demonstrator that provides rapid and cost-effective prototyping and validation of radiofrequency (RF) and microwave (uW) power amplifier (PA) linearization techniques, such as digital predistortion (DPD) and crest factor reduction (CFR). The demonstrator, a part of which is fruit of a close collaboration with UPC University, nowadays comprises a Matlab Testbench including the baseband TX PHY, RX PHY, DPD & CFR blocks, which is able to interface either high-end laboratory instruments or COTS boards (such as pattern generators, A/D, D/A and FPGA/DSP evaluation boards). SHAPER also enables HW/SW co-simulation and validation of DPD and CFR building block implementations in FPGA (on-going work) bridging the gap between applied research and end-product prototyping and validation.
SHAPER can be used to:
- Extract and validate PA behavioral models or characterize the performance of PA designs.
- Design, simulate, validate and experimentally benchmark (through NMSE, ACLR, EVM, BER…) Matlab DPD & CFR algorithms for multiple PA technologies and wireless standards and multi-channel|band|rate configurations.
- Emulate HW constraints (i.e. A/D and D/A bits, data finite precision) before implementation and experimentally evaluate techniques to reduce digital linearization HW complexity & resources.
- Validate product-oriented FPGA prototyping through Matlab + HW/SW cosimulation or real-time full HDL implementations (targeted by the end of 2015).
- Waveform synthesis and acquisition for different BB/BBIQ/IF/RF DUT interfacing modes reaching up to 2 GHz DPD bandwidth. Millimeter wave operation (i.e. E-band) is expected by 2016.
- Key parameter multi-dimensional analysis: i.e. NMSE, ACLR, EVM, rawBER versus Output power, DPD coefficients, PAPR reduction, PCA reduction or any combination of the previous.
- Design of statistically representative experiments and waveforms (Modulated signal full PAPR statistical characterization & waveform design).
The modular Matlab Testbench has been created to be flexible and scalable and thus ease the inclusion of new modems and linearization algorithms, or provide different complexity usage levels. It is also featuring full setup and results traceability and high measurement repeatability.
The benefits brought by PA linearization may have a significant impact on the reduction of the equipment Total Cost of Ownership (TCO) in backhaul radios. Given the fact that small cell densification will be required to address the exponentially growing mobile data traffic demand, the effort put on cost-effective and highly performing radios for small cells backhauling is called to play a determinant role as a key enabler of next generation communications.