Qualcomm Corporate R & D

Mentors: Aidin Bassam, Nick Carbone

June 2018 – September 2018

Smart RF Intern, Corp R&D

Machine learning for Digital Pre-distortion (DPD) to enable transmitter linearity

---

Being part of the Smart RF team was a great experience and gave me perfect exposure to the world of Coorporate R&D. Especially because I was contributing directly to carve out the research direction of the company and hence the industry. It did not hurt that the internship was in San Diego. Amazing place, amazing weather. Truly, one of the best, most productive and happiest summers ever!

Projects

Power amplifiers are pivotal components in any Tx system that have high distortion at high efficiencies. Hence a digital pre-distortion (DPD) block is essential to linearize the output. Classical polynomial series based techniques can no longer meet the DPD spec requirements in the 5G scenario. I developed NN based models that beat classical methods in performance with little overhead in complexity (~3dB improvement in EVM on average, > 6dB best case), over multiple operating conditions as part of the Smart RF team (with amazing mentors).