Pub Date : 2023-12-25DOI: 10.1109/TNANO.2023.3346868
Ying Feng;Mingwei Liang;Ying Li
In this study, the control approach for the piezoelectric actuating system under broad ranges frequency is designed. Addressing the frequency dependent property (rate-dependent) hysteresis of the input signal in piezoelectric actuators, a phenomenological model, a rate-dependent Prandtl-Ishlinskii (RDPI) model, has been utilized to predict the output actuating ability of piezoelectric actuators. Considering the existence of hysteresis and actuator saturation, an adaptive controller with an anti-windup compensator is designed. The actuating performance and the global stability are guaranteed by virtue of the RDPI hysteresis model and the designed anti-saturation filter blocks. The good performance of the proposed control method in mitigating the negative effects under various saturation conditions has been demonstrated by comparing simulation and experimental results.
{"title":"Adaptive Controller With Anti-Windup Compensator for Piezoelectric Micro Actuating Systems","authors":"Ying Feng;Mingwei Liang;Ying Li","doi":"10.1109/TNANO.2023.3346868","DOIUrl":"https://doi.org/10.1109/TNANO.2023.3346868","url":null,"abstract":"In this study, the control approach for the piezoelectric actuating system under broad ranges frequency is designed. Addressing the frequency dependent property (rate-dependent) hysteresis of the input signal in piezoelectric actuators, a phenomenological model, a rate-dependent Prandtl-Ishlinskii (RDPI) model, has been utilized to predict the output actuating ability of piezoelectric actuators. Considering the existence of hysteresis and actuator saturation, an adaptive controller with an anti-windup compensator is designed. The actuating performance and the global stability are guaranteed by virtue of the RDPI hysteresis model and the designed anti-saturation filter blocks. The good performance of the proposed control method in mitigating the negative effects under various saturation conditions has been demonstrated by comparing simulation and experimental results.","PeriodicalId":449,"journal":{"name":"IEEE Transactions on Nanotechnology","volume":"23 ","pages":"45-54"},"PeriodicalIF":2.4,"publicationDate":"2023-12-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"139473633","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Reinforcement, extinction, generalization and differentiation are all basic principles of Pavlov associative memory. Most memristive neural networks that simulate associative memory only consider reinforcement and extinction, while ignoring differentiation and generalization. In this paper, a memristive circuit of associative memory with generalization and differentiation is proposed to solve the above problem. It implements the functions of learning, forgetting, long-term memory, generalization and differentiation. Learning and forgetting correspond to reinforcement and extinction in associative memory respectively. Spontaneous recovery, in which forgotten reflexes can reappear in the absence of an unconditional stimulus, is also discussed here. Besides, a special differentiation method that takes into account the time delay is designed and demonstrated. The proposed memristive circuit of associative memory provides a reference for the theoretical research and application of artificial neural networks.
{"title":"Memristive Circuit Design of Associative Memory With Generalization and Differentiation","authors":"Juntao Han;Xin Cheng;Guangjun Xie;Junwei Sun;Gang Liu;Zhang Zhang","doi":"10.1109/TNANO.2023.3346402","DOIUrl":"https://doi.org/10.1109/TNANO.2023.3346402","url":null,"abstract":"Reinforcement, extinction, generalization and differentiation are all basic principles of Pavlov associative memory. Most memristive neural networks that simulate associative memory only consider reinforcement and extinction, while ignoring differentiation and generalization. In this paper, a memristive circuit of associative memory with generalization and differentiation is proposed to solve the above problem. It implements the functions of learning, forgetting, long-term memory, generalization and differentiation. Learning and forgetting correspond to reinforcement and extinction in associative memory respectively. Spontaneous recovery, in which forgotten reflexes can reappear in the absence of an unconditional stimulus, is also discussed here. Besides, a special differentiation method that takes into account the time delay is designed and demonstrated. The proposed memristive circuit of associative memory provides a reference for the theoretical research and application of artificial neural networks.","PeriodicalId":449,"journal":{"name":"IEEE Transactions on Nanotechnology","volume":"23 ","pages":"35-44"},"PeriodicalIF":2.4,"publicationDate":"2023-12-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"139399700","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2023-12-25DOI: 10.1109/TNANO.2023.3347004
Chiranjib Bhowmick;Sharique Ali Asghar;Pranab Kumar Dutta;Manjunatha Mahadevappa
Breast Cancer remains a devastating affliction for humanity, particularly due to its low survival rates, especially when detected at advanced stages and metastasis has occurred. Early diagnosis is crucial to increase survival rates, but current diagnostic techniques such as mammography and MRI are costly and require an experienced radiologist to interpret results. Transillumination is a non-invasive diagnostic technique that uses light to detect breast abnormalities. LEDs are important in transillumination as the illumination pattern can show abnormalities in breast tissues. Graphene-based materials offer a promising avenue for the creation of thin, flexible, and durable two-dimensional (2-D) light-emitting sources. At the same time, the exceptionally high carrier mobility of graphene ( $sim 15000 frac{text{cm}^{2}}{V sec }$ ) and low band gap in graphene/Si heterojunction LEDs enhances the performance of such LEDs resulting in lower resistance, improved electron-hole recombination, high external quantum efficiency as compared to the traditional GaAs ( $sim 4000 frac{text{cm}^{2}}{V sec }$ ) and AlGaAs ( $sim 212 frac{text{cm}^{2}}{V sec }$ ) LEDs. Hence, the present work explores the potential use of a graphene/silicon hetero junction LED in breast cancer diagnosis. At 690 nm Graphene/Silicon hetero junction LED has a responsivity of $285 frac{rm mA}{rm W}$ whereas responsivity of $0.6 frac {rm{mu A}}{rm{W}}$ and $0.9frac {rm{mu A}} {rm{W}}$ were obtained in the AlGaAs and GaAs LEDs. The present study also investigates the effect of light penetration on breast tissues, including temperature-dependent absorptivity, using the designed LED. The internal quantum efficiency of the proposed LED is also found to be around 37.5% as compared to 24% and 27% for AlGaAs and GaAs LEDs. Thus, the development of such low-cost, radiation-free, and efficient diagnostic optoelectronic devices for breast cancer could significantly increase survival rates, especially in developing countries where access to expensive diagnostic techniques is limited.
乳腺癌对人类来说仍然是一种毁灭性的疾病，特别是由于其存活率低，尤其是在晚期发现并发生转移时。早期诊断是提高存活率的关键，但目前的诊断技术，如乳房 X 射线照相术和核磁共振成像，成本高昂，而且需要经验丰富的放射科医生来解读结果。透射照明是一种非侵入性诊断技术，利用光来检测乳房异常。LED 在透射照明中非常重要，因为照明模式可以显示乳腺组织的异常。石墨烯基材料为制造轻薄、柔韧、耐用的二维（2-D）发光源提供了一条前景广阔的途径。同时，石墨烯超高的载流子迁移率（$sim 15000 frac{text{cm}^{2}}{V sec }$）和石墨烯/硅异质结 LED 的低带隙提高了此类 LED 的性能，从而降低了电阻、与传统的 GaAs ($sim 4000 frac{text{cm}^{2}}{V sec }$) 和 AlGaAs ($sim 212 frac{text{cm}^{2}}{V sec }$) LED 相比，石墨烯/硅异质结 LED 的电阻更低，电子-空穴重组得到改善，外部量子效率更高。因此，本研究探讨了石墨烯/硅异质结 LED 在乳腺癌诊断中的潜在用途。在690纳米波长下，石墨烯/硅异质结LED的响应率为285美元，而响应率分别为0.6美元和0.9美元。在 AlGaAs 和 GaAs LED 中分别得到了 {rm{mu A}}{rm{W}}$ 和 $0.9frac {rm{mu A}}{rm{W}}$ 。本研究还利用所设计的 LED 研究了光穿透对乳腺组织的影响，包括随温度变化的吸收率。研究还发现，拟议 LED 的内部量子效率约为 37.5%，而 AlGaAs 和 GaAs LED 的内部量子效率分别为 24% 和 27%。因此，开发这种低成本、无辐射、高效的乳腺癌诊断光电设备可以显著提高生存率，尤其是在那些昂贵的诊断技术有限的发展中国家。
{"title":"Design and Evaluation of Graphene-Silicon Heterojunction LEDs for Breast Cancer Detection","authors":"Chiranjib Bhowmick;Sharique Ali Asghar;Pranab Kumar Dutta;Manjunatha Mahadevappa","doi":"10.1109/TNANO.2023.3347004","DOIUrl":"https://doi.org/10.1109/TNANO.2023.3347004","url":null,"abstract":"Breast Cancer remains a devastating affliction for humanity, particularly due to its low survival rates, especially when detected at advanced stages and metastasis has occurred. Early diagnosis is crucial to increase survival rates, but current diagnostic techniques such as mammography and MRI are costly and require an experienced radiologist to interpret results. Transillumination is a non-invasive diagnostic technique that uses light to detect breast abnormalities. LEDs are important in transillumination as the illumination pattern can show abnormalities in breast tissues. Graphene-based materials offer a promising avenue for the creation of thin, flexible, and durable two-dimensional (2-D) light-emitting sources. At the same time, the exceptionally high carrier mobility of graphene (\u0000<inline-formula><tex-math>$sim 15000 frac{text{cm}^{2}}{V sec }$</tex-math></inline-formula>\u0000) and low band gap in graphene/Si heterojunction LEDs enhances the performance of such LEDs resulting in lower resistance, improved electron-hole recombination, high external quantum efficiency as compared to the traditional GaAs (\u0000<inline-formula><tex-math>$sim 4000 frac{text{cm}^{2}}{V sec }$</tex-math></inline-formula>\u0000) and AlGaAs (\u0000<inline-formula><tex-math>$sim 212 frac{text{cm}^{2}}{V sec }$</tex-math></inline-formula>\u0000) LEDs. Hence, the present work explores the potential use of a graphene/silicon hetero junction LED in breast cancer diagnosis. At 690 nm Graphene/Silicon hetero junction LED has a responsivity of \u0000<inline-formula><tex-math>$285 frac{rm mA}{rm W}$</tex-math></inline-formula>\u0000 whereas responsivity of \u0000<inline-formula><tex-math>$0.6 frac {rm{mu A}}{rm{W}}$</tex-math></inline-formula>\u0000 and \u0000<inline-formula><tex-math>$0.9frac {rm{mu A}} {rm{W}}$</tex-math></inline-formula>\u0000 were obtained in the AlGaAs and GaAs LEDs. The present study also investigates the effect of light penetration on breast tissues, including temperature-dependent absorptivity, using the designed LED. The internal quantum efficiency of the proposed LED is also found to be around 37.5% as compared to 24% and 27% for AlGaAs and GaAs LEDs. Thus, the development of such low-cost, radiation-free, and efficient diagnostic optoelectronic devices for breast cancer could significantly increase survival rates, especially in developing countries where access to expensive diagnostic techniques is limited.","PeriodicalId":449,"journal":{"name":"IEEE Transactions on Nanotechnology","volume":"23 ","pages":"95-101"},"PeriodicalIF":2.4,"publicationDate":"2023-12-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"139473632","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2023-12-25DOI: 10.1109/TNANO.2023.3346945
Amir Khodabakhsh;Amir Amini;Mohammad Fallahnejad
The evolution trend of wireless communication systems tends to ultra-high data rate, ultra-low latency, and high bandwidth systems. It is foreseen that 6G wireless communication systems will be developed in the range of 100–300 GHz (upper mmWave band) and 300–3000 GHz (terahertz band). In such frequencies, the performance of junctionless field effective transistors is limited due to the reduction of carrier mobility in the device channel. In this paper, for the first time, a shell doped device is proposed to improve RF merit parameters and high-frequency noise performance of GaN junctionless double surrounding nanotube FET device with dual material outer gate (SD-GaN-JNFET). Simulation results show that the doping engineering in the proposed device reduces scattering caused by phonon and doping and increases electron mobility significantly. Parameters g mmax and ƒ T of the SD-GaN-JNFET device in channel length of 15 nm are 666 μS and 8.47 THz, respectively, and NF min <0.025>21 = 22.10 dB and NF = 0.032 dB in central band frequency (140 GHz) was attained. This article opens up an opportunity to achieve high-performance LNA for D-Band 6G applications with the reliable SD-GaN-JNFET device.
{"title":"Highly Linear and Low Noise Shell Doped GaN Junctionless Nanotube TeraFET for the Design of Ultra-Wideband LNA in 6G Communications","authors":"Amir Khodabakhsh;Amir Amini;Mohammad Fallahnejad","doi":"10.1109/TNANO.2023.3346945","DOIUrl":"https://doi.org/10.1109/TNANO.2023.3346945","url":null,"abstract":"The evolution trend of wireless communication systems tends to ultra-high data rate, ultra-low latency, and high bandwidth systems. It is foreseen that 6G wireless communication systems will be developed in the range of 100–300 GHz (upper mmWave band) and 300–3000 GHz (terahertz band). In such frequencies, the performance of junctionless field effective transistors is limited due to the reduction of carrier mobility in the device channel. In this paper, for the first time, a shell doped device is proposed to improve RF merit parameters and high-frequency noise performance of GaN junctionless double surrounding nanotube FET device with dual material outer gate (SD-GaN-JNFET). Simulation results show that the doping engineering in the proposed device reduces scattering caused by phonon and doping and increases electron mobility significantly. Parameters g\u0000<sub>mmax</sub>\u0000 and \u0000<italic>ƒ</i>\u0000<sub>T</sub>\u0000 of the SD-GaN-JNFET device in channel length of 15 nm are 666 μS and 8.47 THz, respectively, and NF\u0000<sub>min</sub>\u0000<0.025>21</sub>\u0000 = 22.10 dB and NF = 0.032 dB in central band frequency (140 GHz) was attained. This article opens up an opportunity to achieve high-performance LNA for D-Band 6G applications with the reliable SD-GaN-JNFET device.","PeriodicalId":449,"journal":{"name":"IEEE Transactions on Nanotechnology","volume":"23 ","pages":"70-77"},"PeriodicalIF":2.4,"publicationDate":"2023-12-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"139406748","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Non-Volatile SRAMs exhibit zero static power loss which is eminent for future on-chip memories. Thus, in this brief, two simple and scalable designs of NVSRAM (8T&9T-2D-2MTJ) having key properties (high speed, low dynamic power, stability and reliability) have been investigated. To make the circuit scalable, the perpendicular magneto anisotropy (PMA) based magnetic tunnel junction (MTJ) (works on the spin transfer torque (STT) & spin orbit torque (SOT) interplay phenomenon) are used as an NV cell in the proposed circuit. A small pulse of current is required for store and restore operation of NV cell, which can effectively accommodate through cross coupled inverters, makes external write circuitry redundant and reduce the complexity of bit cell. The 8T/9T-2MTJ circuit provides an approximate 38%/35% decrease in store delay along with 7%/32% power efficiency as compared to existing NVSRAMs. The hold and write stability of 8T-2MTJ cell is better. However, the read stability of 9T-2MTJ cell is better. The physics-based field-free STT-SOT MTJ model and 40 nm UMC CMOS design kit based NVSRAM make the circuit practically viable. The Process variation analysis and monte carlo analysis suggests that the proposed NVSRAMs are reliable for a wide range (−25 °C to 125 °C) of temperatures and supply voltages. The 8*8 array design of proposed NVSRAMs shows promising results for dense memory architecture.
{"title":"Highly Reliable, Stable, and Store Energy Efficient 8T/9T-2D-2MTJ NVSRAMs","authors":"Sandeep Tripathi;Sudhanshu Choudhary;Prasanna Kumar Misra","doi":"10.1109/TNANO.2023.3345304","DOIUrl":"https://doi.org/10.1109/TNANO.2023.3345304","url":null,"abstract":"Non-Volatile SRAMs exhibit zero static power loss which is eminent for future on-chip memories. Thus, in this brief, two simple and scalable designs of NVSRAM (8T&9T-2D-2MTJ) having key properties (high speed, low dynamic power, stability and reliability) have been investigated. To make the circuit scalable, the perpendicular magneto anisotropy (PMA) based magnetic tunnel junction (MTJ) (works on the spin transfer torque (STT) & spin orbit torque (SOT) interplay phenomenon) are used as an NV cell in the proposed circuit. A small pulse of current is required for store and restore operation of NV cell, which can effectively accommodate through cross coupled inverters, makes external write circuitry redundant and reduce the complexity of bit cell. The 8T/9T-2MTJ circuit provides an approximate 38%/35% decrease in store delay along with 7%/32% power efficiency as compared to existing NVSRAMs. The hold and write stability of 8T-2MTJ cell is better. However, the read stability of 9T-2MTJ cell is better. The physics-based field-free STT-SOT MTJ model and 40 nm UMC CMOS design kit based NVSRAM make the circuit practically viable. The Process variation analysis and monte carlo analysis suggests that the proposed NVSRAMs are reliable for a wide range (−25 °C to 125 °C) of temperatures and supply voltages. The 8*8 array design of proposed NVSRAMs shows promising results for dense memory architecture.","PeriodicalId":449,"journal":{"name":"IEEE Transactions on Nanotechnology","volume":"23 ","pages":"89-94"},"PeriodicalIF":2.4,"publicationDate":"2023-12-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"139473631","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2023-12-18DOI: 10.1109/TNANO.2023.3343667
Harrison Jin;Hanqing Zhu;Keren Zhu;Thomas Leonard;Jaesuk Kwon;Mahshid Alamdar;Kwangseok Kim;Jungsik Park;Naoki Hase;David Z. Pan;Jean Anne C. Incorvia
With the rise in in-memory computing architectures to reduce the compute-memory bottleneck, a new bottleneck is present between analog and digital conversion. Analog content-addressable memories (ACAM) are being recently studied for in-memory computing to efficiently convert between analog and digital signals. Magnetic memory elements such as magnetic tunnel junctions (MTJs) could be useful for ACAM due to their low read/write energy and high endurance, but MTJs are usually restricted to digital values. The spin orbit torque-driven domain wall-magnetic tunnel junction (DW-MTJ) has been recently shown to have multi-bit function. Here, an ACAM circuit is studied that uses two domain wall-magnetic tunnel junctions (DW-MTJs) as the analog storage elements. Prototype DW-MTJ data is input into the magnetic ACAM (MACAM) circuit simulation, showing ternary CAM function. Device-circuit co-design is carried out, showing that 8-10 weight bits are achievable, and that designing asymmetrical spacing of the available DW positions in the device leads to evenly spaced ACAM search bounds. Analyzing available spin orbit torque materials shows platinum provides the largest MACAM search bound while still allowing spin orbit torque domain wall motion, and that the circuit is optimized with minimized MTJ resistance, minimized spin orbit torque material resistance, and maximized tunnel magnetoresistance. These results show the feasibility of using DW-MTJs for MACAM and provide design parameters.
{"title":"Domain Wall-Magnetic Tunnel Junction Analog Content Addressable Memory Using Current and Projected Data","authors":"Harrison Jin;Hanqing Zhu;Keren Zhu;Thomas Leonard;Jaesuk Kwon;Mahshid Alamdar;Kwangseok Kim;Jungsik Park;Naoki Hase;David Z. Pan;Jean Anne C. Incorvia","doi":"10.1109/TNANO.2023.3343667","DOIUrl":"https://doi.org/10.1109/TNANO.2023.3343667","url":null,"abstract":"With the rise in in-memory computing architectures to reduce the compute-memory bottleneck, a new bottleneck is present between analog and digital conversion. Analog content-addressable memories (ACAM) are being recently studied for in-memory computing to efficiently convert between analog and digital signals. Magnetic memory elements such as magnetic tunnel junctions (MTJs) could be useful for ACAM due to their low read/write energy and high endurance, but MTJs are usually restricted to digital values. The spin orbit torque-driven domain wall-magnetic tunnel junction (DW-MTJ) has been recently shown to have multi-bit function. Here, an ACAM circuit is studied that uses two domain wall-magnetic tunnel junctions (DW-MTJs) as the analog storage elements. Prototype DW-MTJ data is input into the magnetic ACAM (MACAM) circuit simulation, showing ternary CAM function. Device-circuit co-design is carried out, showing that 8-10 weight bits are achievable, and that designing asymmetrical spacing of the available DW positions in the device leads to evenly spaced ACAM search bounds. Analyzing available spin orbit torque materials shows platinum provides the largest MACAM search bound while still allowing spin orbit torque domain wall motion, and that the circuit is optimized with minimized MTJ resistance, minimized spin orbit torque material resistance, and maximized tunnel magnetoresistance. These results show the feasibility of using DW-MTJs for MACAM and provide design parameters.","PeriodicalId":449,"journal":{"name":"IEEE Transactions on Nanotechnology","volume":"23 ","pages":"20-28"},"PeriodicalIF":2.4,"publicationDate":"2023-12-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"139090355","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Computing-in-memory (CIM) is a promising candidate for highly energy-efficient neural networks, alleviating the well-known bottleneck in Von Neumann architecture. MRAM has garnered significant attention in the CIM field, providing advantages in terms of non-volatility, high speed, and endurance. However, most existing MRAM-CIM primarily support low-precision operations, which poses a challenge in fulfilling the requirements of complex neural network models for high inference accuracy. To resolve this dilemma, an in-memory Booth Multiplier is proposed with the aim of enhancing the energy efficiency of neural networks performing multi-bit multiply-and-accumulate (MAC) operations. The MRAM array stores the multiplicand, while the multiplier is encoded by a Booth encoder into corresponding control signals, which perform negation and shift operations, reducing half of the partial products and accelerating the overall processing. Simulation results demonstrate at least a 17.3% improvement in energy efficiency compared to the previous in-SRAM counterpart in 8-bit multiplication.
{"title":"A STT-Assisted SOT MRAM-Based In-Memory Booth Multiplier for Neural Network Applications","authors":"Jiayao Wu;Yijiao Wang;Pengxu Wang;Yiming Wang;Weisheng Zhao","doi":"10.1109/TNANO.2023.3343834","DOIUrl":"https://doi.org/10.1109/TNANO.2023.3343834","url":null,"abstract":"Computing-in-memory (CIM) is a promising candidate for highly energy-efficient neural networks, alleviating the well-known bottleneck in Von Neumann architecture. MRAM has garnered significant attention in the CIM field, providing advantages in terms of non-volatility, high speed, and endurance. However, most existing MRAM-CIM primarily support low-precision operations, which poses a challenge in fulfilling the requirements of complex neural network models for high inference accuracy. To resolve this dilemma, an in-memory Booth Multiplier is proposed with the aim of enhancing the energy efficiency of neural networks performing multi-bit multiply-and-accumulate (MAC) operations. The MRAM array stores the multiplicand, while the multiplier is encoded by a Booth encoder into corresponding control signals, which perform negation and shift operations, reducing half of the partial products and accelerating the overall processing. Simulation results demonstrate at least a 17.3% improvement in energy efficiency compared to the previous in-SRAM counterpart in 8-bit multiplication.","PeriodicalId":449,"journal":{"name":"IEEE Transactions on Nanotechnology","volume":"23 ","pages":"29-34"},"PeriodicalIF":2.4,"publicationDate":"2023-12-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"139109643","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
In this article, an electrostatically-actuated NEMS XOR gate is proposed based on photonic crystals for hardware implementation of binary neural networks. The device includes a 2D photonic crystal which is set on a movable electrode to implement the XOR logic using the transmission of specific wavelengths to the output. This design represents the importance of the proposed structure in which the logic gate operation is not dependent on the contact of its conductive layers. Consequently, one of the major issues in MEMS-based logic gates, which is due to the contact of the operating electrodes and may cause stiction problem, reducing the reliability of the system, can be tackled by the present approach. Furthermore, according to the simulation results, the functional characteristics of the present NEMS XOR gate are obtained as follows: pull-in voltage of V p = 8V, operating voltage of V o = 10V and switching time of t s = 4 μs. The results also show that the proposed design provides a classification error rate of between 1% to 12%, while used in neural network implementation. This error can be negligible compared to the state-of-the-art designs in neural network implementation. These appropriate parameters of the present NEMS gate make it a promising choice for the implementation of neural networks with a high network accuracy even in the presence of significant process variations.
{"title":"High Speed Binary Neural Network Hardware Accelerator Relied on Optical NEMS","authors":"Yashar Gholami;Fahimeh Marvi;Romina Ghorbanloo;Mohammad Reza Eslami;Kian Jafari","doi":"10.1109/TNANO.2023.3343618","DOIUrl":"https://doi.org/10.1109/TNANO.2023.3343618","url":null,"abstract":"In this article, an electrostatically-actuated NEMS XOR gate is proposed based on photonic crystals for hardware implementation of binary neural networks. The device includes a 2D photonic crystal which is set on a movable electrode to implement the XOR logic using the transmission of specific wavelengths to the output. This design represents the importance of the proposed structure in which the logic gate operation is not dependent on the contact of its conductive layers. Consequently, one of the major issues in MEMS-based logic gates, which is due to the contact of the operating electrodes and may cause stiction problem, reducing the reliability of the system, can be tackled by the present approach. Furthermore, according to the simulation results, the functional characteristics of the present NEMS XOR gate are obtained as follows: pull-in voltage of V\u0000<sub>p</sub>\u0000 = 8V, operating voltage of V\u0000<sub>o</sub>\u0000 = 10V and switching time of t\u0000<sub>s</sub>\u0000 = 4 μs. The results also show that the proposed design provides a classification error rate of between 1% to 12%, while used in neural network implementation. This error can be negligible compared to the state-of-the-art designs in neural network implementation. These appropriate parameters of the present NEMS gate make it a promising choice for the implementation of neural networks with a high network accuracy even in the presence of significant process variations.","PeriodicalId":449,"journal":{"name":"IEEE Transactions on Nanotechnology","volume":"23 ","pages":"63-69"},"PeriodicalIF":2.4,"publicationDate":"2023-12-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"139399593","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2023-12-14DOI: 10.1109/TNANO.2023.3342844
Junqi Huang;T. Nandha Kumar;Haider A. F. Almurib;Fabrizio Lombardi
Approximate cells can be used to design Ripple Carry Adders (RCAs) for realizing approximate addition in energy-efficient CMOS digital circuits. As inputs of approximate cells could be non-commutative in nature, approximate adders may show different output values under a commutative operation, and this may have a significant effect on the generated sum. This paper presents a detailed analysis of the commutative addition in RCAs made of different approximate cells. Initially, the impact of a non-commutative addition (NCA) to RCAs by approximate cells is assessed by exhaustive simulation at adder level. The results show that at most 17% of additions executed using AFA3 suffer from the non-commutative property, while the values for other adder cells can reach 75%∼99%. Then, an extensive analysis using images from a publicly available library is performed by comparing three-image additions with two-image additions. As a further evaluation, the adders are assessed in an image denoising application. As expected, the effect of NCA is especially pronounced for some non-commutative adders, such as AA2 and AMA4. NCA is also cumulative with the number of approximate additions, thereby causing a significant variation in the output image quality. In terms of metrics, the largest average difference in mean error distance (DMED) for three-image addition is 5.3 times higher than for two-image addition. Rankings of the non-commutative approximate adders show that AMA3 and AFA1 based adders are the best schemes with respect to commutative addition; they both also show good performance in image denoising.
{"title":"On the Commutative Operation of Approximate CMOS Ripple Carry Adders (RCAs)","authors":"Junqi Huang;T. Nandha Kumar;Haider A. F. Almurib;Fabrizio Lombardi","doi":"10.1109/TNANO.2023.3342844","DOIUrl":"https://doi.org/10.1109/TNANO.2023.3342844","url":null,"abstract":"Approximate cells can be used to design Ripple Carry Adders (RCAs) for realizing approximate addition in energy-efficient CMOS digital circuits. As inputs of approximate cells could be non-commutative in nature, approximate adders may show different output values under a commutative operation, and this may have a significant effect on the generated sum. This paper presents a detailed analysis of the commutative addition in RCAs made of different approximate cells. Initially, the impact of a non-commutative addition (NCA) to RCAs by approximate cells is assessed by exhaustive simulation at adder level. The results show that at most 17% of additions executed using AFA3 suffer from the non-commutative property, while the values for other adder cells can reach 75%∼99%. Then, an extensive analysis using images from a publicly available library is performed by comparing three-image additions with two-image additions. As a further evaluation, the adders are assessed in an image denoising application. As expected, the effect of NCA is especially pronounced for some non-commutative adders, such as AA2 and AMA4. NCA is also cumulative with the number of approximate additions, thereby causing a significant variation in the output image quality. In terms of metrics, the largest average difference in mean error distance (DMED) for three-image addition is 5.3 times higher than for two-image addition. Rankings of the non-commutative approximate adders show that AMA3 and AFA1 based adders are the best schemes with respect to commutative addition; they both also show good performance in image denoising.","PeriodicalId":449,"journal":{"name":"IEEE Transactions on Nanotechnology","volume":"23 ","pages":"265-273"},"PeriodicalIF":2.4,"publicationDate":"2023-12-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"140540879","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
The resistive random access memory with metallic carbon nanotube (CNT-RRAM) electrode possesses low power consumption and low junction temperature. In this work, both physical and compact models describing the operations of CNT-RRAM at the microscopic level are presented. In the physics-based model, the migration of oxygen vacancies is described by fully coupled oxygen transport, current continuity, and heat conduction equations, with a proper finite element based numerical solver utilized to solve them. The accuracy of the physical model is verified by comparing the simulated I - V curves with experimental results. After that, a 1T1R memory cell architecture composing of the CNT-RRAM and a vertical MOSFET switch is developed, and a compact model is proposed to characterize its electric properties. The I - V curves obtained by the compact model agree well with experimental data. The results indicate that the proposed models can accurately account for the set/reset characteristics of CNT-RRAM, which would be beneficial for the optimal design of devices and circuits.
{"title":"Modeling and Simulation of RRAM With Carbon Nanotube Electrode","authors":"Da-Wei Wang;Jia-He Zhu;Yi-Fan Liu;Gaofeng Wang;Wen-Sheng Zhao","doi":"10.1109/TNANO.2023.3341414","DOIUrl":"https://doi.org/10.1109/TNANO.2023.3341414","url":null,"abstract":"The resistive random access memory with metallic carbon nanotube (CNT-RRAM) electrode possesses low power consumption and low junction temperature. In this work, both physical and compact models describing the operations of CNT-RRAM at the microscopic level are presented. In the physics-based model, the migration of oxygen vacancies is described by fully coupled oxygen transport, current continuity, and heat conduction equations, with a proper finite element based numerical solver utilized to solve them. The accuracy of the physical model is verified by comparing the simulated \u0000<italic>I</i>\u0000-\u0000<italic>V</i>\u0000 curves with experimental results. After that, a 1T1R memory cell architecture composing of the CNT-RRAM and a vertical MOSFET switch is developed, and a compact model is proposed to characterize its electric properties. The \u0000<italic>I</i>\u0000-\u0000<italic>V</i>\u0000 curves obtained by the compact model agree well with experimental data. The results indicate that the proposed models can accurately account for the \u0000<sc>set/reset</small>\u0000 characteristics of CNT-RRAM, which would be beneficial for the optimal design of devices and circuits.","PeriodicalId":449,"journal":{"name":"IEEE Transactions on Nanotechnology","volume":"23 ","pages":"1-8"},"PeriodicalIF":2.4,"publicationDate":"2023-12-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"139034158","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：