/AI/ML Wireless Technology Recognition Using Photonic-Electronic Neuromorphic Engine

AI/ML Wireless Technology Recognition Using Photonic-Electronic Neuromorphic Engine

Gent | More than two weeks ago

Energy-efficient AI at the Edge with imec's integrated photonic / electronic technologies
Massive MIMO is the most compelling sub-6 GHz physical layer technology for providing uniformly good service to wireless terminals in high-mobility environments. 5G New Radio Unlicensed (NR-U) massive MIMO operates in the same frequency band as WiFi. Due to this coexistence, intermittent interference can lead to performance degradation in individual technologies. Addressing this challenge has led to recent proposals involving complex GPU-based AI/ML classification methods. While these approaches have potential, they suffer from extended training periods and elevated latency, which can impact real-time responsiveness.

Simultaneously, there's a burgeoning field of research exploring the integration of MIMO wireless signals with fiber-optic networks, resulting in a hybrid infrastructure that capitalizes on the strengths of both technologies. Specifically, Radio over Fiber (RoF) links between the Baseband Unit (BBU) and Remote Radio Head (RRH) offer advantages such as high bandwidth, low latency, and long-range connectivity. MIMO wireless technology complements this by extending network coverage, improving mobility support, and facilitating last-mile connectivity. Notably, analog RoF configurations centralize various signal processing functions, leading to cost reductions in remote antenna units.

Leveraging the potential synergy between MIMO wireless signals and RoF technology presents a promising opportunity to integrate edge neuromorphic computing with electronic-photonic integrated circuits (EPICs). Identifying suitable technologies at both the BBU and RRH levels is crucial for making informed spectrum management decisions. Ensuring compliance with the stringent requirements of 5G networks necessitates swift execution of signal detection in the fronthaul. Here, the utilization of photonic-electronic neuromorphic engine architectures proves essential, as they offer rapid and power efficient detection capabilities.

Required background: Electronics Engineering, Photonics and AI/ML expertise will be an added advantage.

Type of work: 80% design/modeling/simulation/, 10% experimental and 10% literature

Supervisor: Xin Yin

Co-supervisor: Adnan Shahid

Daily advisor: Peter De Heyn

The reference code for this position is 2024-086. Mention this reference code on your application form.

Who we are
Accept marketing-cookies to view this content.
Cookie settings
imec's cleanroom
Accept marketing-cookies to view this content.
Cookie settings

Send this job to your email