The Technology
Analog-to-Digital converters are core components in all applications that require translation of analog signals into digital representation.
Traditional ADCs and fixed-rate samplers are constrained by high power consumption due to high-frequency sampling, inefficient data encoding for signals with varying amplitudes or sparse events, and performance degradation at low signal-to-noise ratios or limited bandwidths.
The new Adaptive Integrate-and-Fire Time Encoding Machine (AIF-TEM) is a novel asynchronous analog-to-digital conversion (ADC) technology that converts continuous analog signals into digital time-encoded events. Unlike traditional converters that sample amplitude at uniform intervals, AIF-TEM captures information through non-uniform time events that depend on the signal’s amplitude and frequency.
AIF-TEM introduces a real-time adaptive bias mechanism that continuously adjusts its sensitivity according to the input signal’s characteristics, optimizing both sampling density and quantization accuracy. This adaptive capability allows the system to maintain high fidelity while dramatically reducing energy and data requirements.
Advantages
- Low-Power and Clock-Free Operation: Fully asynchronous design eliminates the need for a global clock, reducing power consumption by orders of magnitude – ideal for portable or battery-operated systems.
- Superior Efficiency: Achieves up to 12 dB lower sampling distortion and 14 dB lower quantization error than conventional Integrate-and-Fire TEMs (IF-TEMs).
- Bit-Rate Reduction: Provides the same reconstruction quality using less than 30 % of the bits required by non-adaptive methods.
- Dynamic Quantization: Introduces a new quantization strategy that adapts step size to signal amplitude, offering an additional 10 dB improvement in reconstruction quality.
- System Cost and Complexity: Extremely low-power operation enables reducing the cooling and power supply requirements, allowing size and cost minimization of high-data-rate systems
Applications
- Ultra-Low-Power Sensing: Internet-of-Things (IoT) and wearable sensors requiring continuous monitoring with minimal power drain.
- Biomedical Devices: Efficient encoding of EEG, ECG, and neural signals in implantable or portable health-monitoring systems.
- Edge AI and Neuromorphic Computing: Event-driven data encoding compatible with spiking neural network architectures.
- Wireless Communication: Bandwidth-efficient signal transmission and reconstruction in in power sensitive systems.
Business Development Contacts