Author: admin

A Music Composition for my Daughter

admin

My daughter has been showing some interest in violin since her grandmother gave her a violin on her 4th birthday. By the time she turned 5, she could play a few pieces. She was very happy when she could play for her music school’s concert last year. Her class teacher also encouraged her to play in front of the class. Some of the children fidgeted. When someone started to say, “I can play too, I have a bigger violin at home”, she became very shy and stopped playing. During the lockdown she played xylophone instead. I composed this song for her when we were walking in the park, and told her that she should not listen to other people’s taunts and be discouraged. Unfortunately, we don’t always get an appreciative audience when we are learning. She is always happier when she could walk in the park and hear the chirping of the birds. So the trills in the music emulate the chirping of the birds. I hope this music will motivate her to continue playing violin.

A Walk in the Park by Louisa Cheong
Acknowledgement: Lenneke Nijst from House of Music, for violin accompaniment
CNN

CNN

admin

Cognitive Radio Modulation and Coding Schemes using Convolutional Neural Networks

This research manifests a process of design, implementation and testing of a deep learning model based on Convolutional Neural Networks (CNN), with the objective to investigate transfer learning, its hyper parameters and its effects on the CNN’s cognitive ability to classify radio modulation and coding schemes. The interest in Deep Learning architecture such as CNN is fuelled by the advent of general purpose Graphical Processing Units (GPUs). NVIDIA’s CUDA parallel computing platform enables accelerated computing intensive application running on thousands of GPU cores in parallel. It is hoped that the high success of CNN in image classification task enabled by the faster computational throughput can be emulated in the modulated radio signal classification task. The classification task provides awareness of modulation and coding rate used by nearby emitters, which can be utilised for Dynamic Spectrum Access (DSA) to minimise interference and optimise spectrum allocation. The radio signals with different modulation and coding rate are generated for WLAN802.1ac and LTE in Matlab. The CNNs in varying model architecture are used for 2 tasks :

  • the source task- classify modulation and coding rate in WLAN802.1ac, and
  • the target task – classify modulation and coding rate in LTE

It is discovered that the CNN acquires high performance when classifying radio signals in LTE, at a level similar to that obtained by source task. The accuracy is higher than when CNN trains on only LTE radio signals. When the dataset is enhanced via “borrowing” and increasing dataset to threefold its size, the performance is higher for both source and target tasks. Using a CNN in transfer learning, i.e. training it on source task, and using it for target task yields prediction accuracy up to 77.54%, which is 32.41% more than without transfer learning. Further adjustment to increase similarity of source dataset to target dataset yielded prediction accuracy of up to 79.34%

This paper was published at the Digital Image and Signal Processing (DISP-19) conference held at the University of Oxford, United Kingdom.

LoRaWAN: Comparison of LoRa Classes and their Power Consumption (October 2016)

LoRaWAN: Comparison of LoRa Classes and their Power Consumption (October 2016)

admin

The quest for low-powered and long-range connectivity in the context of Internet of Things (IoT) is ubiquitous. The technology to bridge the low-powered requirements of a connected building scenario and the long range of cellular networks is seen in LoRaWAN. The paper highlights an experiment conducted at Bell Labs. The objective of the experiment is twofold- to verify the published current levels of different operating modes in datasheet and to compare the battery lifetime for the LoRa Class A and C. A high-end current sensing circuit is constructed to gather the voltage levels and temporal variation with increasing payload sizes and spreading factors. Using the Ohmic Law, the battery lifetime and the energy drain could be calculated and compared across the different classes. As predicted the battery lifetime is shorter for Class B than Class A. The energy drain is the greatest for Class C followed by Class B and Class A. It is also observed that the Sleep Mode of Class A displayed a variation of 3 order-of-magnitudes larger than the benchmark values. This could be evidenced due to the underestimated Sleep values used in the publications. This warrants further investigation. Future works can also be directed to the software development, to access the control of the LoRa device using Host Controller Interface via C/C++.

The paper can be found at https://site.ieee.org/benelux-comvt/scvt-2017-conference-program/

Hello world!

admin

I decided to have a web presence. Not for social reasons (goodbye Facebook), but more for professional reason.

I used to like Facebook, but the fun factor of social media dwindles, as it becomes more invasive with the continuous notifications, news feeds and the advertisements, not to mention that the intermingling of business, family and friends’ contacts has rendered every post to be potentially offensive or a spam.

This website would track my journey as a Data Scientist and to share knowledge that are relevant to the field. If this website has helped you in anyway, please feel free to name and link this website. The number of ingoing and/or outgoing links will increase my ranking on some search engines and will drive traffic to this site. Thank you in advance!