Neuromorphic implementation of a spiking neural network (SNN) for optimizing water supply based on moisture data

alternate text

Current irrigation and water management systems often fail to dynamically adjust to changing environmental conditions, leading to over- or under-watering. Conventional machine learning models that could address this problem require significant computational power, making them less suitable for real-time, edge-device applications in water management. A neuromorphic system based on SNNs can address these challenges by leveraging real-time sensor data in an energy-efficient manner.

The research will be conducted at the Institute of Neuroinformatics (INI, Y55, Irchel Campus ETH/UZH), in the Neuromorphic Cognitive Systems (NCS) group of Prof. Dr. Giacomo Indiveri. The work will be supervised by Dr. Chiara De Luca. Support from the eLIONS Group in Politecnico di Torino will be provided concerning the used dataset.

 

Motivation

Water management is critical to agriculture, especially with increasing climate change and water scarcity concerns. Traditional water supply systems often rely on fixed schedules or basic sensor feedback to control irrigation or water distribution, which can result in inefficiencies and wastage. With the rapid advancements in machine learning and neuromorphic computing, there is an opportunity to develop smarter, more adaptive water supply systems. Spiking Neural Networks (SNNs) on a hardware substrate[1], inspired by the brain's efficient way of processing information, offer a promising solution due to their event-based processing and low energy consumption.

This project proposes the development of a neuromorphic implementation of a Spiking Neural Network on DYNP-SE1[2] to optimize water supply using real-time moisture datasets [3]. The SNN will learn and adapt to environmental changes, ensuring that water is supplied only when necessary, reducing waste, and optimizing water usage.

 

Approach

Taking advantage of the current technologies available in the NCS group, we will implement the feedback-control algorithm on the DYNAP-SE chip [1].

  1. 1. Getting hands-on experience with the DYNAP-SE chip, and a winner-take-all attractor network.

  2. 2. Start-end of water supply time detected with a threshold crossing methods.

  3. 3. Taking into consideration watering schedule constraints and further optimization of the algorithm implementing an on-the-edge moving threshold

  4. 4. Implementing local fault detection mechanisms. (optional)

This project can be tailored on both a semester project and a master thesis project.


Your profile

  • - Interest in Computational Neuroscience (Neuroinformatics), Neuroscience, and Neuromorphic Engineering.

  • - Basic programming skills (Python strongly preferred).

  • - Basic understanding of neural dynamics and dynamical systems.

  • - Interest in testing real-world (electronic) spiking neural networks in real-world (noisy/variable) conditions.


References

[1] Mitra, S., Fusi, S., & Indiveri, G. (2008). Real-time classification of complex patterns using spike-based learning in neuromorphic VLSI. IEEE transactions on biomedical circuits and systems, 3(1), 32-42.

[2] Moradi, S., Qiao, N., Stefanini, F., & Indiveri, G. (2017). A scalable multicore architecture with heterogeneous memory structures for dynamic neuromorphic asynchronous processors (DYNAPs). IEEE transactions on biomedical circuits and systems, 12(1), 106-122.
[3] Canone, Davide, et al. Wappfruit, a project for the optimization of the irrigation in agriculture: final results. No. EGU24-18026. Copernicus Meetings, 2024.

back
back