�ݺ�ߣ

On Problems, Products and
IoT
Dr. Tej Pochiraju, 9th January 2016
micrograce

Brief Bio
● Wireless hardware + applications since 2004
● >150 SMEs + some LEs (Cadbury, Texon, Rockwool)
● ~40 product developments
○ Some successes, many failures
○ Livestock RFID (Scotland/France)
○ Wave energy sensor network (currently off coast of NY)
○ Passive RFID Pressure Sensor (England)
○ Wireless Smart Grid Sensors (Italy)
○ Wireless Tyre Pressure Sensor (Northern Ireland)
○ Trailer Monitoring (sold to Ministry of Defence, UK)
○ Very, very keen to work on environmental challenges in India

Ignore your idea
Definitely ignore your technology
Focus on the problem

Problem Driven Approach to Product Development
● What is the problem?
● Who is affected by this problem?
● Why is this problem worth solving? aka Return on
Investment
● When is the problem worth solving?
● When will the solution be commoditised? Can you wait?
● Who is willing to pay for the solution?
● Is it a product?

Remote Monitoring of Wave Energy Generators

Remote Monitoring of Wave Energy Generators
● NEED: Generator can be damaged if not shut down during strong waves
resulting in loss of ~$1M to equipment owner
● Idea: Remotely monitor waves and automatically shut generator down
during ‘events’
● Technology: Wave sensors in wireless mesh network for remote
monitoring from up to 12 km
● ROI: Sensor Deployment costs about $300,000 and lasts 10+ years [Think
Insurance]

Remote Monitoring of Electric Grid: Similar Story

Touchscreen Sphygmomanometer in 2007!

What is IoT?
● Really old concept
● Overused phrase
● Lots of data but NOT Big Data
○ Most processes being measured are well understood physically or need relatively simple
statistical models
○ Big Data Analytics are great at understanding stochastic/random processes
● Quite useful, in the right context

What is IoT good for?
● Consumer goods - extremely difficult to predict (+ no personal experience)
● Industrial:
○ Where cost of failure is high
■ Asset tracking, predictive maintenance
○ Where process is not completely understood or influenced by uncontrolled parameters
■ Agriculture
■ Chocolate/Alcohol making (haven’t changed for 100+ years!)
■ But beware questions on ROI and who pays!
○ Where there is an advantage to aggregating data from multiple installations
■ Aircrafts (practising data sharing for 60+ years)
■ Traffic monitoring

What I have learnt
● The best technology is one that solves the problem effectively and will solve it
for the lifetime of deployment
● Lifetime determined by first component to fail expensively:
○ Sensor
○ Data logger
○ Communications network (what if 2G disappears?)
● The best way to develop hardware is by eliminating as much hardware as
possible
● If you can implement something in software (and cloud), DO IT!
○ Total Harmonic Distortion calculation on cloud vs in hardware

Common Startup Failures
● Not understanding/picking the right problem
○ Secondary research is not good enough. Get out there and research the problem
● Solving everything
○ Solve one problem really well
■ Kickass cloud db+viz+analytics >> ‘Full-stack’ data logger+cloud :-(
● Lack of Focus
○ Pick a problem that is worth solving, solve it, sell it
○ Do nothing else until 1st sale then whore yourself to investors to scale
○ Set a time limit on when to stop (12-18 months)
● Reinventing the wheel
○ Use open source, use proprietary, use what exists if it meets your needs
○ Replace when worth replacing (almost never will be)

�ݺ�ߣ

On Problems, Products and IoT by Tej | CuTech Talks

More Related Content

On Problems, Products and IoT by Tej | CuTech Talks