Mary Mehrnoosh Eshaghian-Wilner

A nano-scale reconfigurable mesh with spin waves

In this paper, we present a nano-scale reconfigurable mesh that is interconnected with ferromagnetic spin-wave buses. The architecture described here, while requiring the same number of switches and buses as the standard reconfigurable meshes, is capable of simultaneously transmitting N waves on each of the spin-wave buses. Because of this highly parallel feature, very fast and fault-tolerant algorithms can be designed. Furthermore, unlike the traditional spin-based nano structures, which transmit charge, here waves are transmitted. As a result of this, the power consumption of the proposed modules may be low. And using phase logic, simple operations such as AND/OR/NOT can be performed efficiently on the transmitted waves.

Emulation of Neural Networks on a Nanoscale Architecture

In this paper, we propose using a nanoscale spin-wave-based architecture for implementing neural networks. We show that this architecture can efficiently realize highly interconnected neural network models such as the Hopfield model. In our proposed architecture, no point-to-point interconnection is required, so unlike standard VLSI design, no fan-in/fan-out constraint limits the interconnectivity. Using spin-waves, each neuron could broadcast to all other neurons simultaneously and similarly a neuron could concurrently receive and process multiple data. Therefore in this architecture, the total weighted sum to each neuron can be computed by the sum of the values from all the incoming waves to that neuron. In addition, using the superposition property of waves, this computation can be done in O(1) time, and neurons can update their states quite rapidly.

Efficient parallel processing with spin-wave nanoarchitectures

In this paper, we study the algorithm design aspects of three newly developed spin-wave architectures. The architectures are capable of simultaneously transmitting multiple signals using different frequencies, and allow for concurrent read/write operations. Using such features, we show a number of parallel and fault-tolerant routing schemes and introduce a set of generic parallel processing techniques that can be used for design of fast algorithms on these spin-wave architectures. We also present a set of application examples to illustrate the operation of the proposed generic parallel techniques.

Nanoscale modules with full spin-wave interconnectivity

In this paper, we present two nanoscale architectures with full spin-wave interconnectivity. The first architecture is a fully interconnected cluster in which each node can simultaneously broadcast to all other nodes, and can concurrently receive and process multiple data. The second architecture is a crossbar interconnected with ferromagnetic buses that, while requiring the same number of switches and buses as the standard crossbar, is capable of simultaneously transmitting multiple waves at different frequencies on each of the spin-wave buses. The significance of these designs is that the communication between the nodes can be done in constant time, which is a noteworthy improvement considering the Ω(log N) lower-bound on the time delay for implementing such networks in VLSI using traditional electrical interconnects. In these architectures, unlike traditional spin-based architectures that transmit charge, the information is encoded into the phase of spin waves. As a result of this, the presented nanoscale designs may have low power consumption.

The spin-wave nanoscale reconfigurable mesh and the labeling problem

In this article, we present a nanoscale reconfigurable mesh which is interconnected by ferromagnetic spin-wave buses. In this architecture, unlike the traditional spin-based nano structures which transmit charge, waves are transmitted. As a result, the power consumption of the proposed modules can be low. This reconfigurable mesh, while requiring the same number of switches and buses as the standard reconfigurable mesh, is capable of simultaneously transmitting N waves on each of the spin-wave buses. Because of this highly parallel feature, very fast and fault-tolerant algorithms can be designed. To illustrate the superior performance of the proposed spin-wave reconfigurable mesh, we present three fast labeling algorithms.

A Nano-Scale Crossbar with Spin Waves

In this paper, we present a nano-scale crossbar that is constructed with ferromagnetic spin waves. Unlike the traditional spin-based nano structures, which transmit charge, here waves are transmitted. As a result of this, the power consumption of the proposed module may be low. The crossbar architecture described here, while requiring the same number of switches as standard crossbars, is capable of simultaneously transmitting multiple waves on each of the spin-wave paths using different frequencies. As compared to the known molecular crossbars, this design is fault tolerant because alternate paths can be reconfigured in case of a failure on any of the switches.

A Glance at VLSI Optical Interconnects: from the Abstract Modelings of the 1980s to Today’s MEMS Implementations (A Survey Report)

This chapter presents a brief overview of some of the major research contributions in the area of VLSI computing with optical interconnects from the early modelings of the 1980s to today’s MEMS implementations. Both free-space and fiber-guided interconnects are covered. Various models and architectures with optical interconnects are shown, and aspects of their algorithmic design are also reviewed. The chapter concludes with a brief discussion of some of the current advancements in MEMS and nanotechnology that could pave the way towards the actual implementation of some of the theoretical models that were proposed in the 1980s, and eventually towards designing of all optical systems. The materials presented in this chapter are compiled from some of the references that are listed chronologically at the end of the chapter.

Graph Formations of Partial-Order Multiple-Sequence Alignments Using Nanoscale, Microscale, and Multiscale Reconfigurable Meshes

In this paper, we show how to form partial-order multiple-sequence alignment graphs on two types of reconfigurable mesh architectures. The first reconfigurable mesh is a standard microscale that uses electrical interconnects, while the second type of reconfigurable mesh can be implemented at a nanoscale level and employs spin waves for interconnectivity. We consider graph formations for two cases. In one case, the number of distinct variables in the data sequences is constant. In the other case, it can be as much as O(N). We show that given O(N) aligned sequences of length L, we can combine the sequences to form a graph in O(1) time, using either architecture if there is a constant number of distinct variables in the sequence. Otherwise, it will take O(1) time if we use the spin-wave model and O(N) time if we use the standard very large scale integration version.