CSR:Small: High Data Density Short Range Wireless Telemetry for Next Generation IoT Applications

面向下一代物联网（IoT）应用的高数据密度短距离无线遥测（SWiT-IoT）这个SWiT-IoT项目的长期目标是建立一个关于高数据密度短距离无线遥测和节能计算平台的变革性和多学科的研究和创新，该平台集成了定制的模拟射频本发明涉及用于物联网（IoT）应用的（模拟RF）电路和系统架构。无线传感器和传感器网络的激增使物联网应用广泛部署多传感器成为可能。随着无线传感器数量的增加，大量传感器积累的数据量给实时数据处理和有限的无线频谱数据传输带来了瓶颈。为了缓解这些问题，SWiT-IoT研究项目将研究一种两阶段的方法。在第一阶段，该项目将研究一个本地处理或计算单元，用于现场低级别信号处理，数据简化，并在指定带宽内实现大量数据通信。在第二阶段，该项目将研究一种新的脉冲编码方案，使用正交脉冲来进一步压缩数据量。这种双重数据缩减（本地处理和脉冲编码）可在指定带宽内实现高数据速率支持。然后，压缩数据流将被馈送到注入锁定功率振荡器，以驱动天线进行无线传输。拟议的变革性大容量无线数据采集方案将影响各种传感器相关的应用和决策过程，如交通，公共卫生，皮层映射，智能家居等教育目标是整合代表性不足的K-12学生的艺术技能，通过一个有趣的爱好环境激励，参与和帮助学习STEM材料，并让他们为未来的STEM职业做好准备。该项目将继续培训本科生和研究生，举办夏令营，并通过“It's Electric”，“C3-STEM”和UAB CORD计划推广外展计划，以增加高等教育中代表性不足和少数民族的人数。为了实现SWiT-IoT项目的研究目标，主要有三个研究目标：（i）用于数据缩减的能量高效本地处理，（ii）多阶正交脉冲发生器的能量高效模拟RF芯片级实现，以及（iii）基于频谱高效正交脉冲的模拟脉冲序列编码。首先，一个本地处理单元将采用振荡神经网络（ONN）和改进的尖峰检测算法进行研究。该网络利用超低功耗自振荡节点，通过节点间的同步或离散来完成类关联模式识别任务，只传递识别指标同步频率和收敛时间的值，而不传递原始数据。一种改进的基于非线性能量算子的能量有效尖峰检测算法将被研究用于检测信号事件，例如冲击载荷、动作电位、突变等，并中继事件戳以减少数据量。其次，SWiT-IoT项目将研究一种基于修正Hermite多项式的多阶正交脉冲产生方案。将采用创新的神经启发架构，降低系统复杂性，提高能效，并以更小的外形尺寸实现集成电路级实现。第三，一种新的组合脉冲序列编码将允许同时进行多通道无线遥测，具有极好的频谱效率数据密度和数据安全性。在该方案中，将使用n个不同的多阶正交脉冲来为每个信道创建脉冲序列。对于每个信道使用（n-1）个冗余正交脉冲将使得能够支持大量（= n！）信道和（n-1）！-该奖项反映了NSF的法定使命，并通过使用基金会的知识价值和更广泛的影响审查标准进行评估，被认为值得支持。

High Data Density Short Range Wireless Telemetry for Next Generation Internet-of-Things (IoT) Applications (SWiT-IoT)The long-term goal of this SWiT-IoT project is to establish a transformative and multidisciplinary research and innovation on high data-density short-range wireless telemetry and energy-efficient computational platform integrated with custom analog radio-frequency (analog-RF) circuit and system architectures for Internet-of-Things (IoT) applications. Proliferation of wireless sensors and sensor network has enabled widespread deployment of multi-sensors for IoT applications. With the increase of wireless sensors, the cumulative data volume from a large group of sensors is creating bottleneck for real-time data processing and data transmission through a limited wireless spectrum. To alleviate these problems, the SWiT-IoT research project is going to investigate a two-phase approach. On the first phase, the project will investigate a local processing or computational unit for in-situ low-level signal processing, data reduction, and enable high volume of data communication within the specified bandwidth. On the second phase, the project will investigate a novel pulse encoding scheme using orthogonal pulses to compress the data volume further. This two-fold data reduction (local processing and pulse encoding) enables high data-rate support within the specified bandwidth. The compressed data stream will then be fed to an injection-locked power oscillator to drive the antenna for wireless transmission. The proposed transformative large volume wireless data acquisition scheme will influence various sensor related applications and decision-making processes, such as transportation, public health, cortical mapping, smart homes, etc. The education goal is to integrate the artistic skills of underrepresented K-12 students to motivate, engage and help learning STEM materials through a fun loving environment and get them prepared for future STEM careers. The project will continue training undergraduate and graduate students, conduct summer camps, and promote outreach programs through "It's Electric", "C3-STEM" and UAB CORD program, to increase the number of underrepresented and minorities towards higher education.In pursue of the research objectives in the SWiT-IoT project, mainly three research goals are targeted: (i) Energy-Efficient Local Processing for Data Reduction, (ii) An energy-efficient analog-RF chip level implementation of multi-order orthogonal pulse generator, and (iii) Spectral efficient orthogonal pulse based analog pulse-sequence encoding. First, a local processing unit will be investigated by employing oscillatory neural network (ONN) and improved spike detector algorithms. By utilizing ultra-low-power self-oscillating nodes, the ONN will perform class-associative pattern recognition task by synchrony or desynchrony among the nodes, and only relay the values of the recognition indicators frequency of synchronization and convergence time, instead of the raw data. A modified nonlinear energy operator based energy-efficient spike detector algorithm will be investigated for detecting signal events e.g. impact loading, action potentials, abrupt changes, etc. and relay the event stamp to reduce the data volume. Second, the SWiT-IoT project will investigate a Modified Hermite Polynomial based multi-order orthogonal pulse generation scheme. An innovative neuro-inspired architecture will be utilized with reduced system complexity, better energy-efficiency, and integrated circuit level implementation with smaller form factor. Third, a novel combinatorial pulse-sequence encoding will allow simultaneous multichannel wireless telemetry with superb spectrum-efficient data density and data security. In this scheme, n distinct multi-order orthogonal pulses will be used to create pulse-sequence for each channel. The use of (n-1) redundant orthogonal pulses for each channel will enable supporting a large number (= n!) of channels and (n-1)!-times of data rate improvement.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.