Emerging device based spiking neural network (SNN) hardware design has been actively studied. Especially, energy and area efficient synapse crossbar has been of particular interest, but processing units for weight summations in synapse crossbar are still a main bottleneck for energy and area efficient hardware design. In this paper, we propose an efficient SNN architecture with stochastic spin-orbit torque (SOT) device based multi-bit synapses. First, we present SOT device based synapse array using modified gray code. The modified gray code based synapse needs only N devices to represent 2^N levels of synapse weights. Accumulative spike technique is also adopted in the proposed synapse array, to improve ADC utilization and reduce the number of neuron updates. In addition, we propose hardware friendly algorithmic techniques to improve classification accuracies as well as energy efficiencies. Non-spike depression based stochastic spike-timing-dependent plasticity is used to reduce the overlapping input representation and classification error. Early read termination is also employed to reduce energy consumption by turning off less associated neurons. The proposed SNN processor has been implemented using 65nm CMOS process, and it shows 90% classification accuracy in MNIST dataset consuming 0.78J/image (training) and 0.23J/image (inference) of energy with an area of 1.12mm2.

Emerging device-based digital processing-in-memory (PIM) architectures have been actively studied due to their energy and area efficiency derived from analog to digital converter (ADC)-less PIM hardware. However, digital PIM architectures generally need large extra memories to copy parameters, and they also suffer from low computation per memory-cycle efficiencies. In this paper, we present a novel spin-orbit torque magnetic random access memory (SOT-MRAM) based digital PIM architecture to alleviate the extra memory size burden and computation cycle issues. First, we propose the spintronics-assisted logic-in-memory (SLIM) cells to support efficient digital logic operations inside memories, where the voltage-controlled magnetic anisotropy (VCMA) is exploited to enhance the computation per memory-cycle efficiencies. In addition, crossed input source PIM (CRISP) architecture is proposed to extend the merits of SLIM cells by eliminating the extra memories for parameter copying while significantly improving the degree of parallel processing. An intra-memory pipelining scheme is also considered to further increase the throughput of CRISP. The proposed CRISP architecture has been implemented using 28 nm CMOS process, and it presents 1.10 TOPS/W and 0.95 TOPS/mm2, showing considerable improvements of energy efficiency and throughput per area, compared to the state-of-the-art digital PIM architecture. Finally, to evaluate the impact of computation errors induced from the SOT devices and circuits in CRISP architecture, classification accuracy simulations have been performed while applying computation errors.

Spin-transfer torques can switch magnetizations via a current passing through a magnetic tunnel junction, an effect that is being pursued as the switching mechanism in spin-transfer torque magnetic random access memory. Three-terminal devices are also possible. One mechanism is to have a free layer that contains a domain wall that can be manipulated by spin-transfer torques and moved between two configurations that can be read by a separate connection. An alternate approach uses the recent development of spin-orbit torques, which offer an efficient way of manipulating the magnetization of a tunnel junction by current passing through an adjacent layer. These torques allow for the separation of reading and writing currents through three-terminal devices structures. This paper presents the basic principles of spin-orbit torques, the distinguishing features of spin-orbit-torque-induced magnetization dynamics as compared to magnetization dynamics driven by conventional spin-transfer torques. From the application point of view, it presents the pros and cons of spin-orbit-torque-based three-terminal devices including magnetic random access memories. Then, it discusses domain-wall-based three-terminal devices and the advantages and disadvantages of each.

Policy-based network management (PBNM) in the mobile ad-hoc networks (MANETs) requires additional reliable and efficient mechanism over PBNM in wired networks. Thus, it is important that the management system in MANETs should cluster the moving nodes and manage their movements in an effective manner. In this paper, we propose a mechanism for the policy-based management in ad hoc networks in which we discuss methods to discover the policy decision point (PDP), set the management area, and manage the movements of nodes in the PBNM system. Moreover, COPS-PR is extended for the mechanism. Finally, we discuss the results through simulations