Skip to content

Recent Articles


PhD Thesis Defence Passed!

Today at 10am I had my thesis defence. The 4 members or my supervisory committee attended in addition to an internal examiner (from SFU) and external examiner (from UBC). They all attended in person. I had a 20 minute presentation followed by questions. The whole process took about 3 hours. I received very positive feedback for my written thesis and I only have to do minor revisions.

A 5 years journey will soon come to an end. I am hoping that I can continue my load disaggregation, nonintrusive load monitoring or NILM, research. I would like to thank all those who have helped me with my studies.

I will have more to say in the next coming days, stay tuned…


PhD Thesis Submitted for Examination

Well this morning at 9:00am I submitted my PhD Thesis for examination. Apparently the external examiner will write a report after reading it to determine if it is ready for the oral examination/defence. Here is the title, abstract, and keywords:


     Understanding how the electrical appliances and devices in a house consume power is an important factor that can allow occupants to make intelligent and informed decisions about conserving energy — this is not an easy task. These electrical loads can turn ON and OFF either by the actions of occupants, or by automatic sensing and actuation (e.g. thermostat). Even if we could keep track of when loads turned ON and OFF, it is difficult to understand how much a load consumes at any given operational state because of the lack of proper measurement reporting within equipment manuals. Occupants could buy sensors that would help, but this comes at a high financial cost. Power utility companies around the world are now replacing old electro-mechanical meters with digital meters (smart meters) that have enhanced communication capabilities.  These smart meters are essentially a free sensor that offer an opportunity to use computation to infer what loads are currently running in a house and to estimate how much each load is consuming, a process often referred to as load disaggregation.

     This thesis presents a new load disaggregation algorithm (i.e. a disaggregator). This new disaggregator uses a super-state hidden Markov model and a new Viterbi algorithm variant which preserves dependencies between loads and can disaggregate multi-state loads, all while performing computationally efficient exact inference. Our sparse Viterbi algorithm can efficiently compute sparse matrices with a large amount of super-states. Our disaggregator is the first of its kind to run in real-time on an inexpensive embedded processor using low sampling rates (e.g. per minute). Other contributions of the research include an analysis of electrical measurements, the release of a publicly available dataset, and a method for comprehensive accuracy testing and reporting.

Keywords: nonintrusive load monitoring; NILM; energy modelling; hidden Markov model; Viterbi algorithm; sustainability


IEEE Signal Processing Society Summer School: Machine Learning for Big Data

I will be attending the IEEE Signal Processing Society Summer School on Signal Processing and Machine Learning for Big Data. It will be held at UBC from July 29, 2014 to August 1, 2014. From the website:

Humans, machines and sensors collectively generate an enormous amount of data on a daily basis. The fact that much of this data is now accessible provides an opportunity to explore, analyze and extract previously unavailable and potentially highly useful information. In many cases, the volume and speed of data generation makes traditional centralized data analysis infeasible. The lack of structure, and the amount of noise and outliers emphasize the need for robust processing across heterogeneous data domains. High dimensionality makes it challenging to visualize and interpret the data. Overall, Big Data analysis presents many challenges and opportunities for current and future signal processing professionals. This Summer School is intended to provide an introduction to the current efforts to explore Big Data from a signal processing perspective. Topics will range from foundations for Big Data analysis and processing (robust statistical methods, sparse representations, numerical linear algebra, machine learning, convergence and complexity analysis) to Big Data applications (social networks, behaviour and language analysis, bioinformatics, smart grid, environmental monitoring, and others).

Registration deadline is July 15, 2014.
Hope to see you there…

Back from NILM 2014

Well I am back from NILM Workshop 2014. I have a very successful presentation of my disaggregator. I have posted then slide on my NILM website next to where you can download the paper.  Over 90 people attended. There where 8 paper presentations (out of 23 submissions), and 14 poster presentations. My paper was 1 of only 3 papers selected as advancements in NILM algorithm design. I met a number of people. Although our research community is small we mesh well and had very good discussions. BIG thanks to Mario and Zico for organizing it and for Pecan Street’s support.


A children’s book: Puddles or Lunch?

If you have a child who loves dinosaurs and puddles this might be the book for them! Anna has written and illustrated this fun and entertaining book that children will love to read or have read to them. It is available through Kickstarter — a great way to support independent authors who create high quality work. Click on this short link and show your support: There are many great reward to choose from. Here is the book cover and a quote:


“Uh-oh, dinosaurs with muddy feet don’t get lunches! And this dinosaur is hungry! A fun story with a subtle message of resilience.”

A short URL for easy sharing:


Love kickstarter



A Consumer Bill of Rights for Energy Conservation

I just learnt that the main paper I summated to IHTC was accepted titled “A Consumer Bill of Rights for Energy Conservation”. Co-authors were: Laura Guzman Flores, Robyn Gill, Roger Alex Clapp, Lyn Bartram, and Bob Gill. So that is 2 papers accepted at the 2014 IEEE Canada International Humanitarian Technology Conference (IHTC) in Montreal, Canada from June 1-4, 2014. Here is the paper abstract:

Sustainable energy supply and demand can partially be solved by the conservation of energy, which is a personal and self-driven action. However, energy conservation currently requires the purchase of third-party products. The upfront cost of purchasing these products to monitor energy consumption in a home is a barrier that further cements the divide of those that have and those that have not. Detailed appliance power consumption reporting should be made available for free as part of the home’s smart meter. Governments and power utilities must improve and expand policies that promote a socio-economic balance allowing everyone to participate in energy conservation regardless of their economic situation in a sustained way. We critically look at what economics and government polices exist and need to exist. We also demonstrate the computational means to achieve this — nonintrusive load monitoring (NILM) — and discuss how manufacturing and standards organizations need to work together to provide the essential information that describes how appliances consume energy. This paper proposes a Consumer Bill of Rights for Energy Conservation.

Keywords: energy policy, energy economics, smart meter, NILM, load disaggregation, appliance manufactures, standards

We received very good reviews from the reviewers, only some minor edits to do.

I have now posted a PDF of the paper for download.


Transmitting Patient Vitals Over a Reliable ZigBee Mesh Network

I just learnt that a paper that I helped co-author (with Reza Filsoof, Alison Bodine, Bob Gill, and Robert Nicholson) paper “Transmitting Patient Vitals Over a Reliable ZigBee Mesh Network” was accepted at 2014 IEEE Canada International Humanitarian Technology Conference (IHTC) in Montreal, Canada from June 1-4, 2014. Here is the paper abstract:

Real-time measurements and display of vitals are integral part of patients’ health monitoring with limited resources. Wireless sensor networks that communicate in a mesh acquire and transmit such critical parameters, making a medical environment efficient. With limited resources, both equipment and medical staff, the system must be reliable, and capable of dynamic updating on patient status. System must be able to sustain at least single contingency. We designed a signal conditioning circuit and firmware that converts finger pulse into beats per minute (BPM), which is then displayed on a remote station dynamically via ZigBee. Testing was performed to measure the reliability of transmitting data over a wide range of distances through high traffic areas (both human and Wi-Fi). We show that such systems can provided enhanced reliability in a limited resourced environment and have the potential to be deployed in disaster recovery situations.

Keywords: medical, wireless, heart rate, monitoring

This should be a very good conference to attend.


Efficient Sparse Matrix Processing for Nonintrusive Load Monitoring (NILM)

Yesterday I learnt that my full co-authored (with co-authors Ivan Bajic and Fred Popowich) paper “Efficient Sparse Matrix Processing for Nonintrusive Load Monitoring (NILM)” was accepted at NILM Workshop 2014. So I am headed off to Austin, Texas at the beginning of June. Here is the paper abstract:

Nonintrusive load monitoring (NILM) is a process of discerning what appliances are running within a house from processing the power or current signal of a smart meter. Since appliance states are not observed directly, hidden Markov models (HMM) are a natural choice for modelling NILM appliances. However, because the number of HMM states grows rapidly with the number of appliances and their internal states, existing methods have relied on either simplifying the model (e.g. factorial HMM) or reducing the number of appliance states (e.g. considering only on/off states), all of which reduces the accuracy of NILM. In this paper we present a new NILM algorithm that is able to handle multiple internal appliance states while still keeping complexity in check by utilizing the sparsity of HMM emission and transition matrices. The results show overall accuracy results of 95% for both classification and estimation using the AMPds dataset.

Keywords: Load disaggregation, sparse matrix, super-state HMM, Viterbi algorithm

I received very good reviews from the reviewers. I just have some minor edits to do.

I have now posted a PDF of the paper for download.


UBC ICICS: Rethinking Sustainability Symposium 2014

The Rethinking Sustainability Symposium is just around the corner. For the website:

Two Days of Inspiration!

24 – 25 April 2014, Vancouver

This symposium features a broad range of stimulating expert speakers who are “rethinking sustainability.” The underlying philosophy is that trying to change human behaviour to achieve our sustainability goals is not enough. Instead, we should focus on developing new technologies that are both environmentally friendly and improve our quality of life.

Highlighted at the symposium will be a large-scale multidisciplinary approach to sustainability: the ICICS/TELUS People & Planet Friendly Home. This research initiative embraces the symposium philosophy by developing a suite of new technologies that are user-friendly across multiple generations, affordable, and allow for seamless transition as technologies evolve.


See you there!


NILM: What an algorithm can tell you about your energy consumption

Last night I attended the 2014 IEEE Vancouver Section Annual General Meeting. What an amazing event! Leo Del Castillo the General Manager of Xbox Devices at Microsoft was the keynote speaker talking about Xbox One. The Kinect sensor is amazing!

My poster was displayed last night at the Poster Session. I tried with success to make the poster interactive and engaging. Attached to the poster is a cover that hides the NIML disaggregation chart of appliances (see poster 1). The poster was laminated so that the smart meter data chart could be marked using dry erase markers. Once the audience members finished trying to guess what appliances turned ON/OFF and when, they could open the cover to view the NILM chart (poster 2). The audience members could then compare the marked-up smart meter data chart to the NILM chart.

During the session it was very hard for audience members to guess what appliances turned ON/OFF and when. A grounp of 2 people were able to guess that the heating and kettle were ON just before 6:00pm. The poster can be downloaded here.

Keywords: load disaggregation, appliance, smart meter, energy conservation



Get every new post delivered to your Inbox.

Join 55 other followers

%d bloggers like this: