Making Big Data Small

Usage Based Insurance

Insurance companies are increasingly deploying OBD-II dongles and Telematics units, in order to identify different driver profiles, which helps them in personalizing insurance premiums. This means they can better assess the risk that a driver poses onto other traffic participants and identify the exact reason of an accident or an anomaly. The correctness of these models can only be ensured if the amount of data and the variety of car sensors is high enough.

How Teraki can help

In order to make sure that this large volume of data is still cheap and accessible, Teraki has developed patented solutions for efficient acquisition of any kind of automotive sensor data,

• More insight: Needed in order to fairly assess a driver and the road conditions
• Reduced operational costs: Without any increase in costs of mobile data transmission, or additional investments in processing power in the car
• Added Security: By an intrinsic encryption layer of Teraki’s methodology, third party interception of sensitive data is made less likely, key for the handling of privacy affected data

Smartphone Telematics

Driver Analytics via Smartphone Applications: Data collected from the sensors of a smartphone while driving, proves to be a very reliable source for creating driver profiles. And because it doesn’t require direct access to the CAN-bus read-out device, customer adoption is much higher than for dongle solutions.

Current Problems

• Battery usage: GPS data can’t be sampled at high enough frequency, because it drains the battery very quickly
• Cost of mobile data: for accurate information, 2GB of data a month need to be collected and sent from the smartphone sensors
• CPU drainage by compression solutions, leaving less room for applications

Teraki Solution

Our sensor data reduction solution allows developers of automotive applications to reduce the volume of this data by an average 90%. Teraki is allowing them to sample data at much higher rates than what was previously possible with sampling/compression solutions.

What this means:

• Smartphone Battery: Lifetime Improvements of 40-60% in comparison to other compression techniques
• Data cost: Cost of sending out the collected sensor data down with 80%
• CPU allocation: 50% better CPU allocation than with other compression solutions

Telematics Units

Telematics units manufacturers, supplying to automotive OEMs face an important dilemma: Cost pressure from OEMs are reducing their margins, yet they are required to invest into more powerful hardware: they have higher processing for an increased amount of data or alternatively increased in-vehicle storage requirements.

Current Problems

Global players in the field are seeking solutions to solve this issue, but there are restrictions:

• Low computational availability: A small fraction of the total capacity of a CPU is available. Mostly single-core CPUs need to be deployed due to reduced margins
• Restricted reduction rates: The highest possible reduction rate is required on the generated sensor data, without compromising the quality of the reconstruction
• Video processing: Video data needs to be reduced heavily without investing equally heavily into higher performance DSP or GPU hardware

Teraki Solution

Teraki addresses these topics due to its innovative data processing approach: average reduction rates of 90% with an error rate below 1% are reached. The data rates can be adapted on the accuracy thresholds required by the customer while maintaining the lowest CPU and memory requirements for their embedded devices.

This means:

• Continuous mobile communication: Increased average bandwidth is enabled due to 90% average data reduction
• Fine granular insight: Average data rates reach up to several 100Hz per data stream
• Higher margins: Manufacturers can respond faster to the growing data needs of their customers, without sacrificing margins
• Increased Security: The transmission of sensitive data is more secure against third party intervention: Only the party owning the decoder is able to recompose the original data set
• Stable requirements: The embedded implementation in the car is enabled, where stable CPU requirements are critical

Connected Cars

Currently, a car is generating around 60-80 GB of Data per year. Around 250 millions may be connected by 2020 (gartner.com). The number of sensors inside every vehicle will increase from about 200 to 600 in the next 5 year. Sending continuously information will generate a significant cost related with transmission and storage. By reducing the data by up to 99% and an average of 90%, Teraki will support car companies to enable new affordable services including breakthrough innovations like driver sleeping detection. The autonomous car will be data driven by definition.