In this paper, we investigate the granularity in the free layer of the magnetic tunnel junctions (MTJ) and its potential to function as a reservoir for reservoir computing where grains act as oscillatory neurons while the device is in the vortex state. The input of the reservoir is applied in the form of a magnetic field which can pin the vortex core into different grains of the device in the magnetic vortex state. The oscillation frequency and MTJ resistance vary across different grains in a non-linear fashion making them great candidates to be served as the reservoir's outputs for classification objectives. Hence, we propose an experimentally validated area-efficient single granular vortex spin-torque nano oscillator (GV-STNO) device in which pinning sites work as random reservoirs that can emulate neuronal functions. We harness the nonlinear oscillation frequency and resistance exhibited by the vortex core granular pinning of the GV-STNO reservoir computing system to demonstrate waveform classification.
Neuromorphic computing is a promising strategy to overcome fundamental limitations, such as enormous power consumption, by massive parallel data processing, similar to the brain. Here we demonstrate a proof-of-principle implementation of the weighted spin torque nano-oscillator (WSTNO) as a programmable building block for the next-generation neuromorphic computing systems (NCS). The WSTNO is a spintronic circuit composed of two spintronic devices made of magnetic tunnel junctions (MTJs): non-volatile magnetic memories acting as synapses and non-linear spin torque nano-oscillator (STNO) acting as a neuron. The non-linear output based on the weighted sum of the inputs is demonstrated using three MTJs. The STNO shows an output power above 3 µW and frequencies of 240 MHz. Both MTJ types are fabricated from a multifunctional MTJ stack in a single fabrication process, which reduces the footprint, is compatible with monolithic integration on top of CMOS technology and paves ways to fabricate more complex neuromorphic computing systems.
26 July 2016 in IEEE Transactions on Very Large Scale Integration (VLSI) Systems
Spin-transfer torque random access memory (STT-RAM) has emerged as an attractive candidate for future nonvolatile memories. It advantages the benefits of current state-of-the-art memories including high-speed read operation (of static RAM), high density (of dynamic RAM), and nonvolatility (of flash memories). However, the write operation in the 1T-1MTJ STT-RAM bitcell is asymmetric and stochastic, which leads to high energy consumption and long latency. In this paper, a new write assist technique is proposed to terminate the write operation immediately after switching takes place in the magnetic tunneling junction (MTJ). As a result, both the write time and write energy consumption of 1T-1MTJ bitcells improves. Moreover, the proposed write assist technique leads to an error-free write operation. The simulation results using a 65-nm CMOS access transistor and a 40-nm MTJ technology confirm that the proposed write assist technique results in three orders of magnitude improvement in bit error rate compared with the best existing techniques. Moreover, the proposed write assist technique leads to 81% energy saving compared with a cell without write assist and adds only 9.6% area overhead to a 16-kbit STT-RAM array.
Recent research provides examples of neuromorphic systems applied to process biological signals or to interface biological tissues. Usually, in such contexts, the neuromorphic system is used for automatic anomaly detection. The automation of the long-term monitoring of biological signals holds promise for lightening the burden placed on clinicians. At the same time, the adoption of such devices potentially allows processing to be performed locally, without the need to transfer data to an external processor. In turn, on-site signal analysis makes closed-loop intervention feasible, to correct the source of the anomalies. So far, the common approach has been to deploy the network of spiking neurons either on multi-core neuromorphic platforms or on programmable units (field programmable gate array). However, if the aim is to develop wearable or even chronically implantable devices, it is imperative to move in the direction of embedded solutions, tailored and optimized for the specific application. To this end, the present study proposes a neuromorphic device implemented in CMOS technology for the detection of epileptic seizures (ictal events) from local field potential (LFP) signals. The LFP data have been acquired by a multi-electrode array in a slice of mouse hippocampus-cortex. The system includes an analog-to-event converter (AEC) encoding the recorded signals into trains of spikes, and a small spiking neural network (SNN) of 2 × 1 neurons with online biological-plausible learning. The AEC yields two spike-trains: UP spikes that account for the positive slope signal and DOWN spikes that correspond to the negative slope signal. The synapses among the input and output layers are plastic and follow the spike-timing-dependent plasticity rule. Early results show that the SNN module is able to detect ictal events with a delay of 64.98 ± 30.92 ms, consuming 50 pW. The layout of the entire system occupies 528 µm × 278 µm.
28 March 2019 in IEEE Transactions on Circuits and Systems II: Express Briefs
In spintronic-based neuromorphic computing systems (NCSs), the switching of magnetic moment in a magnetic tunnel junction (MTJ) is used to mimic biological neuron firing. However, the high currents required for MTJ switching will lead to a large energy consumption and low speed operation of the MTJ-based NCSs. In this paper, the energy consumption and the delay of the proposed MTJ-based NCS are reduced by thermally assisting the MTJ switching through a microwatt nanosecond laser pulse. Simulation results show 85.7%, 84% and 97.8% improvements in energy consumption, delay and energy-delay product, respectively, for the laser-assisted spintronic NCS (LAS-NCS) by heating the MTJs up to 127∘C. The energy consumption of the laser pulse for heating up the MTJ stack is 1.2X-2.4X lower than Joule heating method. Moreover, in contrast with Joule heating method, the LAS-NCS does not require MTJ with modified structure.