SHF: Small: Ferroelectric Transistor based Coupled Oscillators for Non-Boolean Computing

由于基于输入布尔处理的传统计算体系结构对于涉及识别和感觉处理的任务来说可能是次优的，因此探索新的非布尔计算技术具有承担重要角色的潜力。本项目旨在通过研究新的后CMOS器件和电路，并广泛分析它们在应用级别上的影响来解决这一问题。研究成果可能会对计算机辅助诊断、语音/人脸识别、数据分类和资源分配等关键应用产生直接影响，并使医疗保健、国防和安全等多个领域受益。此外，通过建议的技术实现的省电可能会转化为移动系统更长的电池寿命，从而促进更丰富的用户体验和超低功耗处理。该项目将通过将研究成果整合到课程中来加强研究生和本科教育。建议的研究利用铁电晶体管的新兴技术，通过将铁电材料固有的非线性转化为持续的振荡，设计出新型的低功率和紧凑的振荡器。多个这样的振荡器，其动力学是可控制的，被耦合以实现非布尔计算结构。振荡器之间的同步程度提供了关于不同信号之间的匹配/不匹配程度的信息，使得能够针对诸如模式匹配、运动检测等应用进行有效的决策。提出的方法包括广泛的器件和电路的联合设计，以利用铁电晶体管提供的独特功能来提高耦合振荡器的能量和面积效率。研究工作跨越了设计抽象的不同层次，包括(I)探索符合所建议方法的新型铁电器件，(Ii)新的振荡器设计，以及(Iii)全面的应用级分析。将对铁电振荡器进行广泛的基准测试，以对照以前探索过的振荡器，全面量化它们的优势和权衡。此外，还将进行详细的研究，以建立器件-电路特性与铁电材料固有特性之间的关系，以揭示所提出的振荡器的基本属性及其丰富的振荡动力学。

Since conventional computing architectures based on Boolean processing of inputs may be sub-optimal for tasks involving recognition and sensory processing, exploration of new non-Boolean computing technologies has the potential to assume an important role. This project aims at addressing this issue by investigating new post-CMOS devices and circuits, and extensively analyzing their implications at the application level. The research outcomes are likely to have a direct impact on critical applications such as computer aided diagnosis, speech/face recognition, data classification and resource allocation, and benefit several areas such as healthcare, defense, and security. Moreover, power savings achieved with the proposed techniques may translate to longer battery life for mobile systems facilitating a richer user experience and ultra-low power processing. The project will enhance graduate and undergraduate education by integrating the research outputs in the curriculum. The participation of under-represented groups will also be encouraged.The proposed research utilizes the emerging technology of ferroelectric transistor to design novel low power and compact oscillators by translating the inherent non-linearity of the ferroelectrics to sustained oscillations. A multitude of such oscillators, whose dynamics can be controlled, are coupled to realize a non-Boolean computing fabric. The extent of synchronization amongst the oscillators provides information about the degree of match/mismatch amongst different signals, enabling efficient decision-making for applications such as pattern matching, motion sensing and others. The proposed approach involves extensive co-design of devices and circuits to harness the unique features offered by the ferroelectric transistors for enhancing the energy- and area-efficiencies of the coupled oscillators. The research effort spans different levels of design abstraction including (i) the exploration of novel ferroelectric based devices amenable for the proposed approach, (ii) new oscillator designs, and (iii) a comprehensive application-level analysis. Extensive benchmarking of ferroelectric based oscillators will be performed against previously explored oscillators to comprehensively quantify their benefits and trade-offs. In addition, detailed investigation establishing the relationships between the device-circuit characteristics and the inherent properties of the ferroelectric materials will be carried out to uncover the fundamental attributes of the proposed oscillators and their rich oscillation dynamics.