Direct time-of-flight (d-ToF) estimation with high frame rate requires the incorporation of a time-todigital converter (TDC) at pixel level. A feasible approach to a compact implementation of the TDC is to use the multiple phases of a voltage-controlled ring-oscillator (VCRO) for the finest bits. The VCRO becomes central in determining the performance parameters of a d-ToF image sensor. In this paper we are covering the modeling, design and measurement of a CMOS pseudo-differential VCRO. The oscillation frequency, the jitter due to mismatches and noise and the power consumption are analytically evaluated. This design has been incorporated into a 64×64-pixel array. It has been fabricated in a 0.18μm standard CMOS technology. Occupation area is 28×29μm2 and power consumption is 1.17mW at 850MHz. The measured gain of the VCRO is of 477MHz/V with a frequency tuning range of 53%. Moreover, it features a linearity of 99.4% over a wide range of control frequencies, namely from 400MHz to 850MHz. The phase noise is of -102dBc/Hz at 2MHz offset frequency from 850MHz. The influence of these parameters in the performance of the TDC has been measured. The minimum time bin of the TDC is 147ps with a RMS DNL/ INL of 0.13/ 1.7LSB.

Conference - IEEE International Symposium on Circuits and Systems ISCAS 2018

This paper analyzes three cryptographic modules suitable for digital designs of trusted virtual sensors into integrated circuits, using 90-nm CMOS technology. One of them, based on the keyed-hash message authentication code (HMAC) standard employing a PHOTON-80/20/16 lightweight hash function, ensures integrity and authentication of the virtual measurement. The other two, based on CAESAR (the Competition for Authenticated Encryption: Security, Applicability, and Robustness) third-round candidates AEGIS-128 and ASCON-128, ensure also confidentiality. The cryptographic key required is not stored in the sensor but recovered in a configuration operation mode from non-sensitive data stored in the non-volatile memory of the sensor and from the start-up values of the sensor SRAM acting as a Physical Unclonable Function (PUF), thus ensuring that the sensor is not counterfeit. The start-up values of the SRAM are also employed in the configuration operation mode to generate the seed of the nonces that make sensor outputs different and, hence, resistant to replay attacks. The configuration operation mode is slower if using CAESAR candidates because the cryptographic key and nonce have 128 bits instead of the 60 bits of the key and 32 bits of the nonce in HMAC. Configuration takes 416.8 μs working at 50 MHz using HMAC and 426.2 μs using CAESAR candidates. In the other side, the trusted sensing mode is much faster with CAESAR candidates with similar power consumption. Trusted sensing takes 212.62 μs at 50 MHz using HMAC, 0.72 μs using ASCON, and 0.42 μs using AEGIS. AEGIS allows the fastest trusted measurements at the cost of more silicon area, 4.4 times more area than HMAC and 5.4 times more than ASCON. ASCON allows fast measurements with the smallest area occupation. The module implementing ASCON occupies 0.026 mm2 in a 90-nm CMOS technology.

Journal Paper - Microprocessors and Microsystems, vol. 61, pp 108-116, 2018 ELSEVIER
DOI: 10.1016/j.micpro.2018.05.016    ISSN: 0141-9331    » doi

Interest in event-based vision sensors has proliferated in recent years, with innovative technology becoming more accessible to new researchers and highlighting such sensors' potential to enable low-latency sensing at low computational cost. These sensors can outperform frame-based vision sensors regarding data compression, dynamic range, temporal resolution and power efficiency. However, available mature frame-based processing methods by using Artificial Neural Networks (ANNs) surpass Spiking Neural Networks (SNNs) in terms of accuracy of recognition. In this paper, we introduce a Hybrid Neural Network which is an intermediate solution to exploit advantages of both event-based and frame-based processing. We have implemented this network in FPGA and benchmarked its performance by using different event-based versions of MNIST dataset. HDL codes for this project are available for academic purpose upon request.

Conference - IEEE International Symposium on Circuits and Systems ISCAS 2018

The SpiNNaker chip is a multi-core processor optimized for neuromorphic applications. Many SpiNNaker chips are assembled to make a highly parallel million core platform. This system can be used for simulation of a large number of neurons in real-time. SpiNNaker is using a general purpose ARM processor that gives a high amount of flexibility to implement different methods for processing spikes. Various libraries and packages are provided to translate a high-level description of Spiking Neural Networks (SNN) to low-level machine language that can be used in the ARM processors. In this paper, we introduce and compare three different methods to implement this intermediate layer of abstraction. We have examined the advantages of each method by various criteria, which can be useful for professional users to choose between them. All the codes that are used in this paper are available for academic propose.

Conference - IEEE International Symposium on Circuits and Systems ISCAS 2018

Neuromorphic Computing or Spiking (also called Event-Driven) Neural Systems are becoming of high interest as they potentially allow for lower power hardware computing platforms, where power consumption is data driven. Traditional approaches (both in software and in hardware), which are not data driven, rely on generic system state updates, consuming a fixed amount of computing resources at each step, independent on the data itself. In neuromorphic spiking or (event-driven) computing systems power is consumed (in principle) if new data is transferred, either at the system input, system output, or internally between computing nodes. One such neuromorphic event-driven computing platform is the scalable SpiNNaker system, which is aimed for a million ARM core platform, capable of emulating in the order of a billion neurons in real time. An important practical drawback of the platform is the long time it takes to download to the hardware a given computational architecture. This step has to be repeated even if one wants to update a set of parameters. Here we present a method for updating internal parameters without downloading again the full architecture, by adding special neurons into the computing architecture which when they spike change given parameters. This allows to download the computing architecture only once to the SpiNNaker platform, and then take advantage of its highly efficient communication network to command specific parameter changes. This allows for intensive parameter searches in a more efficient manner.

Conference - IEEE International Symposium on Circuits and Systems ISCAS 2018