Joint CFR and DPD optimization for 5G Digital Front Ends

Sammanfattning: This thesis project investigates the topic of joint optimization for crest factor reduction and digital predistortion blocks, which are used for digital correction of analogue impairments in digital radio front ends. The work first investigates the effects of joint parameter variation of the above blocks on the global performance and computational cost of the digital chain, given reference CFR and DPD algorithms. Then it proposes a method for joint optimization of said CFR and DPD parameters, based on a constrained optimization of a composite target function. The constraints of this optimization problem ensure that the solution meets the requirements on out-of-band emissions and signal quality, while the target function values solutions that outperform these requirements and are computationally cheaper to implement. The solutions of such an optimization problem are determined with the help of simulations. The simulations provide estimated performance metrics on signal quality and out-of-band emissions as well as estimated computational cost for a given parameter configuration. The results show that the proposed optimization can guide the selection of parameters for multiple blocks of the DFE while ensuring applicability in different use-cases via an easily configurable target function.