August 2013
Article
AutomatedBuildings.com

Babel Buster Network Gateways: Big Features. Small Price.
Control Solutions, Inc. - Minnesota

(Click Message to Learn More)


Big Data for Smart Buildings

Imagine the power of advanced data analytics using big data architecture to visualize the long term perspectives and true smart devices that perform proactive decisions.

Nirosha MunasingheNirosha Munasinghe
MBusIT BSc BE
(Hons) (Melb)

Contributing Editor

Articles
Interviews
Releases
New Products
Reviews
Control Solutions, Inc
Editorial
Events
Sponsors
Site Search
Newsletters
Securing Buildings News
Archives
Past Issues
Home
Editors
eDucation
ABB
Training
Links
Software
Subscribe
Secured by Cimetrics

As we ride the wave of device connectivity to the cloud delivering a plethora of data, the “Big Data” phenomena is surfacing the market and revolutionizing the way we manage data. New data architecture is forming shifting the paradigm from traditional rationale database management tools.  MapReduce, Hadoop, Chukwa, Ambari, Hive, Zookeeper, Cassendara and Mahout are some of the fundamental architecture driving the concept.  They are transforming data management in the Information Communication Technology (ICT) industry with faster access to data at the right time to visualize to the right audience. Is it time for Big Data for the BAS industry, or can BAS data be analyzed before the big data space, or is there a hybrid solution? This article examines the concept of big data and examines how traditional machine learning techniques can be used in smart devices to outline a hybrid solution to manage data in a building automation system.

Big Data
Figure 1: Big Data concepts surfacing in the market

Big Data describes a set of complex data that is difficult to process using traditional database management tools. The processing includes capturing, storing, searching, sharing, transferring, analyzing and visualization. Big Data is driven by significant growth in data sets over the last few years due to growth in unstructured data from mobile devices, sensing technology, wireless sensors, and social networks, satellite images, photo/video and speech, along with the reduction in cost of storage. The data growth has put limitations on the current rational database technology in obtaining the right data at the right time, which has introduced a new wave of database architecture such as MapReduce and Hadoop. The fundamental advantage of current big data architectures is the ability to process unstructured data. The traditional database tools required structured datasets with relationships to process the data.  However architecture such as Hadoop and MapReduce allow processing of unstructured data at high speed. Hadoop allows big problems to be decomposed into smaller elements so that analysis can be done quickly and cost effectively. It is a versatile, resilient, clustered approach to managing files in big data environment.  This architecture has been a key breakthrough in data management as unstructured data is ever increasing in our day to day life.

Unstructured Information

Figure 2: Unstructured information is everywhere

The new big data architecture has a significant advantage for the BAS industry. The fundamental feature of a smart building is the ability to capture endless data from key performing buildings resources using sensors, wireless and wired networks. The data is transferred to a central processing unit to analyze to take proactive action and then visualize the data for an audience. The rapid growth of data in a building has driven the dashboard industry over the last two years with many players entering the market with various data visualizing panels. As the growth of data increases the big data architecture is a must for the BAS industry to process unstructured data in an efficient manner. However, players need to be careful to avoid the assumption that big data architecture is the only solution in the future for data handling for a smart building. The BAS industry is in a domain where data is captured from key resources to a smart device before it is processed to a database or data warehouse.  The smart device can perform much more functionality than its current capabilities. It can be argued that the word ‘smart’ is the key buzzword that has driven the ICT over the last decade. However are devices really smart? Let’s examine a BAS smart controller. It is currently capturing input data from a sensor network and then outputs action according programmed logic. The data is transferred to a database for further analysis and mining. Is the device really smart?  No, the BAS controller has been performing such functions for over 20 years. Not much has changed in the implementation. Changes are required to take advantage of cheap flash technology available to store the raw data in the device to compute analysis on the fly and take proactive action on the key resources it is monitoring or controlling. Artificial Intelligence technology such as neural networks, fuzzy logic and data mining techniques are proven solutions that can be used in the device to machine learn and make proactive decisions to improve the performance of the resource. For example the most fundamental theory behind a control system is the operation of the PID loop to control an output according to the input and set point that requires user input to tune its parameters. However many integrators or users lack the technical skills to tune these parameters failing its functionality and more importantly wasting the energy resources of a building. The product can have brilliant features but without its core functionality due to user error there is a fundamental problem. A clear solution is to implement an adaptive PID control loop in the device where it learns from past data to adjust its parameter without user intervention. Historical data in the device can be used to achieve such results without the need to send data to a large data warehouse application. Similarly learning applications can be implemented in the device to action energy usage patterns and make proactive decisions on the local device for it to be a truly smart device and a true distributed system. The historical data from the device can be exported to an external application to use big data architecture techniques to perform advance analytics to observe the long term perspective. We have enough capacity and processing power in our smart devices to make short term decisions locally to act faster without being reliant on external applications.

Hybrid Solution
Figure 3: Hybrid Solution to manage data in BAS industry

The vendors in the BAS and energy markets need to plan their road map to big data architecture now. However, most vendors have proven smart devices on their platform which can be used to perform many of the data analysis locally using machine learning techniques and to provide proactive feedback to the energy resources it is monitoring and controlling. Imagine the power of advanced data analytics using big data architecture to visualize the long term perspectives and true smart devices that perform proactive decisions.

footer

Reliable Controls
[Click Banner To Learn More]

[Home Page]  [The Automator]  [About]  [Subscribe ]  [Contact Us]

Events

Want Ads

Our Sponsors

Resources