An Experimental Study of Split-Learning TinyML on Ultra-Low-Power Edge/IoT Nodes

Running deep learning inference directly on ultra-low-power edge/IoT nodes has been limited by the tight memory and compute budgets of microcontrollers. Split learning (SL) addresses this limitation in which it executes part of the inference process on the sensor and off-loads the remainder to a companion device. In the context of constrained devices and the related impact of low-power, over-the-air transport protocols, the performance of split learning remains largely unexplored. TO the best of our knowledge, this paper presents the first end-to-end TinyML + SL testbed built on Espressif ESP32-S3 boards, designed to benchmark the over-the-air performance of split learning TinyML in edge/IoT environments. We benchmark the performance of a MobileNetV2 image recognition model, which is quantized to 8-bit integers, partitioned, and delivered to the nodes via over-the-air updates. The intermediate activations are exchanged through different wireless communication methods: ESP-NOW, BLE, and traditional UDP/IP and TCP/IP, enabling a head-to-head comparison on identical hardware. Measurements show that splitting the model after block_16_project_BN layer generates a 5.66 kB tensor that traverses the link in 3.2 ms, when UDP is used, achieving a steady-state round-trip latency of 5.8 s. ESP-NOW presents the most favorable RTT performance 3.7 s; BLE extends battery life further but increases latency beyond 10s.

Optimizing LoRa for Edge Computing with TinyML Pipeline for Channel Hopping

We propose to integrate long-distance LongRange (LoRa) communication solution for sending the data from IoT to the edge computing system, by taking advantage of its unlicensed nature and the potential for open source implementations that are common in edge computing. We propose a channel hoping optimization model and apply TinyML-based channel hoping model based for LoRa transmissions, as well as experimentally study a fast predictive algorithm to find free channels between edge and IoT devices. In the open source experimental setup that includes LoRa, TinyML and IoT-edge-cloud continuum, we integrate a novel application workflow and cloud-friendly protocol solutions in a case study of plant recommender application that combines concepts of microfarming and urban computing. In a LoRa-optimized edge computing setup, we engineer the application workflow, and apply collaborative filtering and various machine learning algorithms on application data collected to identify and recommend the planting schedule for a specific microfarm in an urban area. In the LoRa experiments, we measure the occurrence of packet loss, RSSI, and SNR, using a random channel hoping scheme to compare with our proposed TinyML method. The results show that it is feasible to use TinyML in microcontrollers for channel hopping, while proving the effectiveness of TinyML in learning to predict the best channel to select for LoRa transmission, and by improving the RSSI by up to 63 %, SNR by up to 44 % in comparison with a random hopping mechanism.

Effective ML Model Versioning in Edge Networks

Machine learning (ML) models, data and software need to be regularly updated whenever essential version updates are released and feasible for integration. This is a basic but most challenging requirement to satisfy in the edge, due to the various system constraints and the major impact that an update can have on robustness and stability. In this paper, we formulate for the first time the ML model versioning optimization problem, and propose effective solutions, including the automation with reinforcement learning (RL) based algorithm. Without loss of generality, we choose the edge network environment due to the known constraints in performance, response time, security, and reliability. The performance study shows that ML model version updates can be fully and effectively automated with reinforcement learning method as compared to other approaches. We show that with a carefully chosen range of traffic load values, the proper versioning can improve the security, reliability and ML model accuracy, while assuring a comparably lower response time.

Towards Smart Microfarming in an Urban Computing Continuum

Microfarming and urban computing have evolved as two distinct sustainability pillars of urban living today. In this paper, we combine these two concepts, while majorly extending them jointly towards novel concepts of smart microfarming and urban computing continuum. Smart microfarming is proposed with applications of artificial intelligence in microfarming, while an urban computing continuum is proposed as a major extension of the concept towards an efficient IoT-edge-cloud continuum. We propose and build a system architecture for a plant recommendation system that uses machine learning at the edge to find, from a pool of given plants, the most suitable ones for a given microfarm using monitored soil values obtained from IoT sensor devices. Moreover, we propose to integrate long-distance LoRa communication solution for sending the data from IoT to the edge system, due to its unlicensed nature and potential for open source implementations. Finally, we propose to integrate open source and less constrained application protocol solutions, such as AMQP and HTTP protocols, for storing the data in the cloud. An experimental setup is used to evaluate and analyze the performance and reliability of the data collection procedure and the quality of the recommendation solution. Furthermore, collaborative filtering is used for the completion of an incomplete information about soils and plants. Finally, various ML algorithms are applied to identify and recommend the optimal plan for a specific microfarm in an urban area.

A Semi-Markov Decision Process -based Channel Allocation Model for Unreliable Terahertz Reconfigurable Intelligent Surfaces

Terahertz (THz) communications and reconfigurable intelligent surfaces (RISs) have been recently proposed to enable various powerful indoor applications, such as wireless virtual reality (VR). For an efficient servicing of VR users, an efficient THz channel allocation solution becomes a necessity. Assuming that RIS component is the most critical one in enabling the service, we investigate the impact of RIS hardware failure on channel allocation performance. To this end, we study a THz network that employs THz operated RISs acting as base stations, servicing VR users. We propose a Semi-Markov decision Process (SMDP)-based channel allocation model to ensure the reliability of THz connection, while maximizing the total long-term expected system reward, considering the system gains, costs of channel utilization, and the penalty of RIS failure. The SMDP-based model of the RIS system is formulated by defining the state space, action space, reward model, and transition probability distribution. We propose an optimal iterative algorithm for channel allocation that decides the next action at each system state. The results show the average reward and VR service blocking probability under different scenarios and with various VR service arrivals and RIS failure rates, as first step towards feasible VR services over unreliable THz RIS.