As smart defence projects are being rolled out, there remains a need to experiment with and to benchmark newer technology in an isolated, smaller and controllable environment, before transiting to large-scale deployment. DSTA proposed the DSTA Integrated Complex as a space for technology incubation and for growing technology-push ideas and capabilities.

The complex served as a natural testbed with an integrated wired and wireless network infrastructure, building sensors and actuators, and a live user-base. It allowed technology to be taken out of the laboratory for testing, before large-scale implementation. By tapping subject matter experts within DSTA for this testbed, engineers who were involved in varied applications of similar new technology could come together to learn as emerging concepts were proven. The DSTA community could also gain from these initiatives, and contribute different perspectives to seed new innovative approaches. The team experimented with various technologies in FY2017.



Video analytics (VA) have demonstrated increased accuracy, with algorithms based on deep learning. The testbed facilitated the benchmarking of various VA algorithms, within an end-to-end system. This means that it covers the entire process of VA, ranging from pre-processing videos to applying machine-learning algorithms on the videos for a desired output.


As such, DSTA engineers were able to assess and experiment with the performance of these VA models.

Moving forward, DSTA will build up more video data-sets that are representative of other deployment settings to enrich the testbed.


Given the variety of sensors installed in an average modern building, the team explored how situational awareness could be improved through insights garnered from the fusion of a myriad of sensor information.


Advancements in commercial technology related to optimising staff experience and security within the complex could also be applied rapidly in this testbed, and be assessed before an organisation-wide roll-out. The technology applied included a trial on the efficiency of security processes such as automated personnel and vehicle access clearance.


There was also potential to uncover energy savings from analysing data generated by sensors that monitored resource utilisation within the complex. The team is applying machine learning techniques on the cooling plant to discern the building’s load profile characteristics and further optimise the parameter settings to reap energy savings.


The end-to-end architecture ranging from sensors to visualisation and IoT platforms has been designed. The IoT platform has been set up in the complex to enable it to be ready for sensor prototyping and IoT initiatives. Condition-based maintenance and a real-time location system for surveillance purposes have been implemented in the complex. Arrangements have also been made with telecommunications providers to extend the base station to the complex in support of narrowband-IoT, thus encouraging collaboration on IoT test methodologies and interoperability. The amount of data collected will make DSTA, MINDEF and the SAF data-enabled and ready for data analytics.