Demonstration of Multivariate Photonics: Blind Dimensionality Reduction With Integrated Photonics

Abstract

Multi-antenna radio front-ends generate a multi-dimensional flood of information, most of which is partially redundant. Redundancy is eliminated by dimensionality reduction, but contemporary digital processing techniques face harsh fundamental tradeoffs when implementing this class of functions. These tradeoffs can be broken in the analog domain, in which the performance of optical technologies greatly exceeds that of electronic counterparts. Here, we present concepts, methods, and a first demonstration of multivariate photonics: a combination of integrated photonic hardware, analog dimensionality reduction, and blind algorithmic techniques. We experimentally demonstrate 2-channel, 1.0 GHz principal component analysis in a photonic weight bank using recently proposed algorithms for synthesizing the multivariate properties of signals to which the receiver is blind. Novel methods are introduced for controlling blindness conditions in a laboratory context. This work provides a foundation for further research in multivariate photonic information processing, which is poised to play a role in future generations of wireless technology.

Photonics for artificial intelligence and neuromorphic computing

“… Machine learning acceleration: vector-matrix multiplications [14], deep learning inference [15], ultrafast or online learning [16]  Intelligent signal processing: wideband RF signal processing [17], fiber-optic communication [18,19] Photonic circuits are well suited to high-performance implementations of neural networks for two predominant reasons: interconnectivity and linear operations. Connections between pairs of artificial neurons are described by a scalar synaptic weight (a primary memory element), so the layout of interconnections can be represented as a matrix-vector operation, where the input to each neuron is the dot product of the output from connected neurons attenuated by a weight vector.…”

Quantifying power use in silicon photonic neural networks

“…Digitization can play several roles in analog photonic information processors. In the linear category, all VMMs require digitization of all channels; field-programmable photonic arrays may require digitization of only a few [50]; and multivariate RF photonics aim to reduce the number of ADCs [51]. Nonlinear photonic neurons based on O/E/O conversion can contain ADCs in order to tailor their transfer function [52], or they can can use electrooptic nonlinearities to avoid digitization altogether [8,53,54].…”

“…The analysis of resolution during correlated signal fan-in in Secs. III are extensible to microwave photonics, including field-programmable photonic arrays [50,64] and multivariate photonics [51,66]. We postulate that most of these systems will exhibit regime-like scaling features that could benefit from visualization strategies in Sec.…”

Photonic analog signal processing and neuromorphic computing [Invited]

“…Their analog nature allows for signal processing in continuous time, and reduces the cost, memory requirements, and precision loss resulting from the need to digitize massive amounts of data. Not only are these advantages for applications involving optical signals, such as optical fiber or free space optical communication [2][3][4][5][6][7][8][9][10][11][12][13] , but they also show promise when applied to computation in general [14][15][16][17][18][19][20][21][22][23][24] .…”

“…They also show promise for accelerating general computation, as many of these devices are capable of being integrated into silicon wafers and used as application-specific co-processors in tandem with digital and analog electrical circuitry [18] . Photonic blind-source separation is a technique that uses photonic circuits to manage interference in wireless communication systems [2][3][4][5] , free-space optical communication systems [4] , and optics-based sensor technologies [29,30] . It takes advantage of the rich amplitude information present in multiinput-multi-output signal processing systems to separate signals that exist in the same electromagnetic bandwidth, circumventing the limitations of traditional frequency filtration and multiplexing techniques.…”