Siva Gawsalyan
My Work Showcase
Motion Capture with Wearables
The platform is developed from scratch using the Madwig AHRS algorithm running real-time on IMU sensors and OpneGL for rendering 3D objects.
​
The captured quaternion information for each limb is transmitted back to the base station (laptop computer) in real-time to produce motion graphics.
​
The code is not made public but there are more videos and one publication link is provided below!
Anomaly detection at the Edge
A novel binary Electrocardiogram (ECG) classification neural network for continuous cardiac observation suitable for a wearable platform to diagnose cardiovascular diseases at their early stages with low computational complexity and power consumption. The presented solution utilizes a novel architecture consisting of Long Short Term Memory (LSTM) cells and Multi-Layer Perceptrons (MLP).
​
The algorithm was mapped to a fixed point implementation, retained, and ported to an embedded platform. In Laboratory testing the overall system design was demonstrated to achieve significant power consumption savings when used to gate the wireless transmission of ECG signals to only broadcast those beats deemed to be anomalous. In addition, the design retains the advantages of having stand-alone continuous cardiac classifications without the need for always-on wireless connectivity making our proposed system very suitable for wearable platforms.
​
Publications will be soon available!
IOT Android App
A demo version of Android app developed for IOT application using MQTT protocol
Get in Touch
Interested in learning more about me, my work or how we can collaborate on an upcoming project? Feel free to reach out anytime, I would be more than happy to chat.
​
Connect with me via
Facebook, Instagram, Twitter, LinkedIn,
GitHub, ResearchGate, and GoogleScholar
"There was no one near to confuse me, so I was forced to become an original"