Embedded Computing: Machine Learning at the Edge

The theory of pushing computing closer to sensors where data is gathered is a significant point of modern embedded computing – i.e., the edge of the network. With machine/deep learning, this conception becomes even more vital to enable autonomy and intelligence at the edge. For many applications such as industrial robots and automated machinery on a factory floor to an agricultural tractor in the field, to self-guided vacuums in the home, the computation must happen locally. There can be various reasons for local processing, such as low latency, reliability, bandwidth, power consumption, and privacy. Quite an amount of applications are driving adoption of machine learning (ML), such as advanced driver assistance systems (ADAS) and self-driving cars, big-data analysis, surveillance, and improving processes from audio noise reduction to natural language processing. Thus the concept of machine learning at the edge is expected to prevail in the future application of embedded computing along with the advancement in IoT concept.

At present, a modern vehicle contains between 25 or 100 electronic control units (ECUs). These systems are generally portioned based on domains, namely real-time body controls and infotainment controls. The real-time body controls include a different category such as chassis control, body control, powertrain control, and active safety control. The second category, the infotainment controls, includes navigation, information management, computing, external communication, and entertainment. Embedded computing in modern vehicles is segmented into different domains, mainly differentiated by the criticality of the function executed. In general, each ECU integrates a processing element (single or multi-core processor), memory subsystems (including volatile and non-volatile), optional dedicated accelerators like cryptographic or image processing engines, power supply elements, and the interfaces to the different sensors, actuators, and network. Specific combinations are chosen depending on the requirements for each application. For example, in the body electronics domain that handles simple comfort functions like doors, access control, lighting systems, and climate control, an ECU architecture may be composed of an 8- to 32-bit micro-controller, non-volatile code memory, and network interfaces like CAN and LIN.

Get sample PDF brochure, Click Here!

Ever-increasing consumer electronics industry in the Asia Pacific is expected to provide a promising growth to the region in the embedded systems market during the forecast period. The region accounts for the major consumer electronics share in the global market, with China being the biggest country. Customer demand for advanced products along with emerging economies in the region is anticipated to offer a sound growth to the Asia Pacific. This factor is escalating the growth of the revenue size of embedded computing market in the Asia Pacific region. This as a result has positively impacted embedded computing market.

The primary reason for the growth of embedded computing market is the The US is a major global leader in technology and associated equipment and controls. Most of the demand for embedded computing comes from the automobile and electronics manufacturing industries. This scope is facilitating the applications to adopt this technology, thereby creating a substantial market for the same at present, which is projecting a positive growth on the embedded computing market. Similarly, The growth in the consumer electronics, automotive electronics, and communication technology is increasing the demand for advanced technologies with an objective to gather real-time information. The country has foreseen the tremendous growth in myriad industries that also includes the transportation equipment industry, food processing industry, and petroleum and coal products industry. Canada has several industrial automation companies which are specialized in operating ICT industries throughout the country.

The major players include in this research are Advantech Co., Ltd., Arms Holdings, Fujitsu, IBM Corporation, Intel Corporation, Microchip Technology Inc., STMicroelectronics, Qualcomm Technologies, Inc., Renesas Electronics Corporation, and Texas Instruments Incorporated. 

Strategic Insights

Some of the market initiatives were observed to be most adopted strategy in the global embedded computing market. Few of the recent market initiatives are listed below:

2019: Advantech announced the introduction of Palm Size Embedded System EPC-U2117, which enhances the CPU performance by 30% as well as boost the graphic performance by 45%.

2018: Intel Corporation partnered with Micron Technology, Inc. to complete joint development for the second generation of 3D XPoint technology.

2017: Intel Corporation launched the world’s first commercially available 64-layer, TLC, 3D NAND solid state drive.

2016: Qualcomm launches Snapdragon 600E and 410E.

Views: 276


You need to be a member of The Engineering Exchange to add comments!

Join The Engineering Exchange


© 2020   Created by Marshall Matheson.   Powered by

Badges  |  Report an Issue  |  Terms of Service