AI (artificial intelligence) is deemed as the last integral part of the IoT (Internet of Things) architecture, and the integration of AI and IoT can generate brand new AIoT business opportunities that can be better tapped with creative thinking. As one of the world's tech hubs, Taiwan has witnessed many businesses create innovative and practical business models associated with AIoT applications by taking advantage of the country's robust IT technological prowess.Among the enterprises, three tech startups presented their latest AIoT solutions at a recent Digitimes-hosted "D Talk" forum: Deep Force, Starwing Technology and AstralNet. They have created new business models with solutions in AI-based image processing, indoor positioning and information security authentication, respectively.Deep Force uses AI to enhance image valuesDeep Force's AI platform can easily help users conduct smart processing of numerous pictures stored in their smartphones for various applications by utilizing deep-learning algorithms, according to company CEO Winston Chen.Chen said that there are many enterprises engaged in the development of AIoT solutions, but their platforms, devices and chips are usually short of good compatibility. In contrast, he stressed, Deep Force's deep-learning algorithms can not only be compatible with different system equipment and components, but also be easily incorporated into enterprises in the same ecosystems to shorten the time needed for product development and launch through its fast, precision, and high-privacy off-line AI technology able to boost the storage capacity.Chen continued that the development of both AI and IoT is driven by vertical applications in diverse fields, and therefore no standardized architectures are available. In demanding customized architecture designs, users usually will set the goals for establishing their application systems, and Deep Force will help them select appropriate neural networks in accordance with the goals and optimize the networks to completely fit the application systems before the neural networks and parameters are incorporated into terminal products.With its AI deep learning technology mainly focused on image processing, Deep Force has rolled out such products as face-unlocking systems and smart albums. The company plans to apply the technology to two-stage authentication, VIP customer services, smart robots, advance driving warning, as well as smart education and sales, which all need AI image processing to boost their service values, according to Chen.For instance, the authentication by facial recognition can make online trading safer, and operators of physical stores can also use the image analysis of security surveillance systems to judge the personal status of clients so as to provide them with more precise service quality. Chen said that images count high in data-based IoT applications, and AI can help derive more values from the images and stimulate different creativity to generate huge business opportunities.Starwing offers centimeter-grade indoor positioning systemOn another front, hospitals or clinics may need to know patients' locations, and supermarkets may want to know the traffic flow of consumers among store shelves. All these could be addressed with indoor positioning solutions now available in the market, but such solutions usually bear a large deviation of several meters, making precision indoor positioning a difficult job. In this regard, Starwing Technology's centimeter-grade wireless positioning technology can be applied to provide much more precise indoor positioning, according to company president Ian Chen.Chen said that the current indoor positioning technology is mainly based on triangulation that requires several devices to perform. This, coupled with indoor decorations and furniture that are likely to disturb signals, will lead to a positioning deviation of 2-3 meters. Starwing's Intelligent Indoor Positioning System (IIPS)-Pro adopts the Bluetooh standard and incorporates machine learning algorithms that can automatically adjust deviations to under 30cm to optimize the positioning effect.In addition, the company has also developed what it calls the world's only 3D vertical positioning technology that can also check the height of the object targeted for positioning, with the deviation also measurable by centimeters.Chen disclosed that Starwing's positioning tags can well resist electronic interferences, and they can be accurately positioned if they are put inside clothing items. In addition, the company's AI-based data analysis engine can also analyze the positioning data as part of big data, and the company has rolled out AI analysis kits including those for tracing control, hotspot analysis, track prediction and advance security warning, among others.Precision indoor positioning can be adopted for multiple and practical applications such as museum indoor guide, analysis of consumer traffic flow at shopping malls, detection of patient locations at hospitals, with the accumulated positioning data able to be analyzed through AI algorithms to work out optimal operating models. The company will move to develop modularized application kits for precision indoor positioning, according to Chen.AstralNet's two-way authentication enhances IoT securityAs to information security, it is an important issue for both general networking and IoT applications. As convenience counts high in IoT applications, how to achieve a balance between security and easy operation for IoT has become the most crucial concern for system developers. Ming-yang Chih, co-founder and CTO of AstralNet, opined that in the AI era, IoT information security must be executed in a smarter, easier and faster way.Chih stressed that AstralNet's two-way SAGE (secure authentication group engine) platform is the exact solution to address the balance issue, as it combines cryptography and AI technology to make IoT designs more secure and easier. Citing an example to highlight the feature of the platform, Chih pointed out that any modern car comes with a key featuring wireless lock control design, which, however, allows only the key to conduct one-way authentication with the locking system whose password cannot be changed automatically. When the key gets lost, anyone picking the key can easily drive away the car even though the driver gets a new key. But AstralNet's SAGE features a dynamic mutual authentication scheme that can generate a new password whenever a new key is used, thus disabling the lost key.Chih continued that AstralNet is composed of professionals in the fields of IoT and cryptography, and their combination has created a distinct encryption technology that can perform easier, securer and faster identity authentication for IoT applications. Two-way authentication can be carried out in many ways. For instance, he indicated, a smartphone and a car key can be connected via Bluetooth to undergo fingerprint recognition before being connected to terminal public key infrastructure for two-way authentication.AstralNet plans to continue expanding the SAGE applications through modularization. Chih said that modularity can accelerate the security designs of IoT systems to facilitate easier and securer device-to-device authentication, stressing that the integration of AI and cryptography can offer smart solutions to eradicate the doubts about security information in IoT applications and promote the popularity of such applications.Speakers and organizers at a recent D Talk forumPhoto: Ambrose Huang, Digitimes, January 2018

The 32 Taiwan startups that showcased products and services at CES 2018 in January attracted 43,211 visitors and held 2,402 business talks with venture capital firms with potential funding and purchases estimated at US$47.62 million and US$604,000 respectively, according to the Ministry of Science and Technology (MOST).To help showcase startups, MOST set up the Taiwan Tech Star pavilion at Eureka Park during CES 2018, a special exhibition zone giving startups the opportunities to launch new products, services or ideas. The Taiwan startups' expertise covers smart medical technology, AR, VR, AI, smart wearable devices and IoT (Internet of Things), MOST noted.Among the startups, Robotelf Technologies won CES 2018 Innovation Award for its home-use robot Robelf; and iXensor won Best of Baby Tech Award for its Eveline ovulation testing system, MOST said.MOST said it will select Taiwan startups for CES 2019 as early as possible to enhance international promotions.MOST will also set up a startup incubator in Taipei and invite local and international accelerators to incubate local and overseas start-up teams.

High performance computing (HPC) will become the most crucial platform in the development of process technologies for AI (artificial intelligence) chips, and CoWoS (chip on wafer on substrate) and SiP (system in package) will emerge as key packaging processes for such chips, according to Digitimes Research.Usually, an AI architecture will include the upstream cloud computing, midstream edge computing and downstream devices. And the performance of AI chips can be boosted by upgrading the microform technology and changing the transistor structure in the front end, or by incorporating advanced packaging technologies in the back end.In the backend packaging, the 2.5D CoWoS process technology launched by Taiwan Semiconductor Manufacturing Company (TSMC) can upgrade the performance of packaged ICs by sharply boosting I/O pin numbers through the incorporation of silicon interposer and the TSV (through si via) technology. In the first half of 2017, TSMC launched an HPC platform using 7nm CoWoS process technology to further better IC performance.In addition, the IoT platform also plays an important role in AI development. As IoT chips involve requirements for low power consumption, low cost and ready availability, SiP will be the main packaging technology applicable to chip solutions for IoT applications.Accordingly, it will be an increasingly important trend for chipmakers to integrate frontend and backend process technologies, Digitimes Research believes, adding that makers must join forces with EDA, IP, and IC designers to build a complete ecosystem if they want to secure a preemptive presence in the AIoT (artificial intelligence IoT) space.

Mainly exported to American and European markets, Cosen's products are representative of the Made in Taiwan vision. Cosen specializes in the production of band saws, which may be unheard of to many people but are fundamental to manufacturing.Transformed Cosen"A band saw to manufacturing is like a knife to cooking. The ingredients have to be chopped before they can be cooked," explained Cosen CEO Alice Wu. The first step in material processing, whether for steel bridge construction, machining or high-precision processing, is usually cutting. In the case of any error in the cutting process, production will not be able to carry on.It is exactly for this reason that manufacturers count on their band saws to stay operable at all times. Furthermore, as the materials in use today improve in quality, band saws are being used to cut harder and harder materials with increasing difficulty. The industry will therefore impose more stringent requirements on band saws, which is also the challenge that Consen has to address now.Industrial big data analytics expert Jie Li once said each country has its unique manufacturing strength and the Taiwan manufacturing industry has to enable its own differentiating value. Extending Li's point of view, Wu commented on what customers really expect from band saws. They need their band saws to precisely cut all the materials in full compliance with the schedule and quality requirements.How to relieve customers from worries of equipment problems after they put Cosen's products to use is the value Cosen strives to provide. Accordingly, Cosen is undergoing transformation from a band saw supplier to a service provider that performs equipment health assessment.According to Cosen, customers' biggest worry is the reliability of the cutting tool on the equipment, which affects the operation of the whole machinery and possibly even the delivery schedule. Customers need to stay aware of equipment conditions and know when the best time to change the cutting tool is. The cutting tool should be replaced before it is completely worn to avoid unexpected downtime.In view of this, Cosen has added an equipment monitoring system in the controller chassis. The system integrates a data capture module along with a signal processing and conversion module and connects to electric current sensors and resistance temperature detectors. Coupled with machine learning and artificial intelligence (AI) technologies, the system not only enables customers to collect signals generated from the band saw during cutting operations but also allows them to build models based on the data and find the pattern that indicates components need to be serviced or replaced.Then, the measurements of the stress level at the saw and the frictional heat generated during cutting are recorded and analyzed such that equipment health and product service life that manufacturers originally had no way of knowing now become clear information.China Steel enhances productivity despite extreme operating conditionsChina Steel is the only steel producer in Taiwan capable of making steel from ore to slabs, blooms and billets with integrated production steps. The process of making steel includes the use of chemical processes, steam/electricity co-generation systems and high-pressure gas. Under these harsh working conditions, equipment operating in a steel mill has to be able to withstand extreme conditions such as high temperature, heavy load and high impact while maintaining high yield.The construction of a steel process plant is a multi-million dollar investment. If any piece of equipment fails, it will seriously affect production not only on the line itself but also upstream and downstream processes. The sophisticated processes from raw material delivery, coking, steel plate and steel bar making to semi-finished steel casting go through a series of massive and intricate pieces of equipment. Their components can easily wear and tear under extreme operating conditions. Without early detection of equipment irregularity, failures are bound to happen and cause equipment shutdown.Due to the magnitude of the equipment, it will take more time to replace components, resulting in longer downtime. For the purpose of minimizing loss due to unexpected equipment shutdown and workplace hazards, China Steel has brought in smart equipment monitoring and diagnostic systems to assess conditions of important production equipment and thereby determine the best time and method to conduct maintenance.For instance, like dough made in a central kitchen of a food factory, semi-finished steel is provided to midstream and downstream steel plants as materials for further processing. After being fed to the furnace for softening, semi-finished steel then goes through the rolling mill to be rolled into various types of steel products. The rolling machine applies a very strong compressive force and one component could easily cost over NT$1 million. Damage or replacement before a component's expected service life incurs a major expense to the manufacturer.Therefore, China Steel has decided to embed sensors in the production equipment to detect the use conditions. The use of equipment monitoring systems enables early discovery of situations such as abrasion resistant plates showing signs of wear and bearings showing irregularity. This is instrumental to maintaining production stability.Machining tool suppliers leverage flutter monitoring to enhance CNC precisionA maker of machining centers located in Taiwan's precision machinery base has a history of over five decades in refined research and production of machining tools. Its products are being exported to many countries around the world and are enthusiastically embraced by internationally renowned companies. The maker is currently stepping up efforts to transition from a traditional machining tool supplier to a world-leading machining tool brand by steering its developments toward smart manufacturing.In addition to continuing development of high-quality models and expanding presence in smart machinery, the machining tool maker also endeavors to optimize machining center functionality by looking into issues its customers have encountered. CNC machinery is critical production equipment to the manufacturing industry so manufacturers constantly look to improve CNC machinery performance in an attempt to boost process precision and product quality.Accurate calibration is crucial to machining precision. Manufacturers have long relied on veteran technicians who have accumulated years of experience to do the calibration. However, such practice is hardly scientific and is also tedious and time-consuming. An oversight in the process could result in loss of precision. Furthermore, veteran technicians will have to retire someday and there is no guarantee they will be able to pass down all their valuable know-hows.Another factor that affects machining quality is flutter. When a CNC tool performs fast or deep cutting, flutter could occur. Flutter is a type of small-amplitude high-frequency vibration which is barely perceptible. However, not only will occurrence of flutter compromise machining precision but it can indirectly cause damage to important components such as cutters, spindles and bearings. In view of this, equipment makers look to detect flutter early in time to stop the problem from becoming so bad that it diminishes machining quality.For the purpose of eliminating uncertainty of manual calibration and minimizing machinery flutter, makers bring in sensor technologies to help their machinery tools conduct self-calibration and real-time monitoring, integrating their machinery with built-in spindle calibration and flutter detection. However, a brand new CNC machining tool may cost anywhere between hundreds of thousands of NT dollars and millions of NT dollars. Redesigning such machinery to add monitoring capability will definitely run up costs.As such, the maker tried to add monitoring modules to existing machining tools, augmenting them with spindle calibration and flutter detection capabilities, to accelerate time-to-market for new CNC machining tools. By selecting a plug-and-play data capture module to complete an easy-to-integrate solution, the maker only had to make small modifications to the hardware circuit, rather than a design overhaul, to complement their machinery with the desired capabilities.

With advancing Industry 4.0 developments, large-scale enterprises generally have no problem putting equipment monitoring in place, which can be handled simply by their own automation departments. However, to small-scale manufacturers not as resourceful, implementing equipment monitoring can be challenging.Predictive and preventive maintenance - an important concept of equipment monitoring - is essential to Industry 4.0 manufacturing. It prevents production shutdown due to unexpected equipment failure or extra costs incurred from unnecessary early replacement of good parts.Large-scale manufacturers, such as TSMC, had already put the concept of predictive and preventive maintenance into practice long before equipment monitoring was widely embraced. They have devoted R&D resources and amassed considerable knowledge and expertise in this respect.The market is already seeing a wide variety of equipment monitoring solutions. The availability of total solutions integrating hardware and software implementations has lowered the threshold for manufacturers to introduce equipment monitoring systems at their factories. Advantech's equipment condition monitoring software WebAccess/MCM, NEXCOM's predictive maintenance system solution and NI's machine condition monitoring (MCM) solution are examples of complete services combining the installation of hardware and software systems and the provision of professional analyses.There are generally two types of manufacturers that have needs for equipment monitoring systems - large enterprises with strong technical capabilitie, and businesses with limited resources. In-house engineers of large-scale companies bring in equipment monitoring systems mainly for the purpose of accelerating development cycle. Less resourceful businesses only now trying to catch up with the equipment monitoring trend do not have employees with the required system development skills or specialized data analytic expertise.Industry 4.0 needs vertical integrations between upstream and downstream industries. An attempt to introduce any new technology can be a whole new challenge to the manufacturer making the move. In view of this, equipment monitoring solution providers have developed ready-to-use software packages to help manufacturers significantly cut down R&D and labor costs. Alternatively, they provide user-friendly graphical interface so that users can quickly learn the system and be able to focus more efforts on studying the signals.Shen-Wei Wu, product manager, Industrial IoT Group, Advantech, indicated a total solution allows users not specialized in equipment data collection and analysis to also be able to put a basic monitoring solution in place without problem. This saves time in the process of installing the system. Some software packages even come with built-in basic data compilation and diagnostic functions, such as time domain and frequency analyses as well as filter data averaging. This is exactly what businesses with no data analysis experts need.On-site experts play a critical roleA total equipment monitoring solution lowers the technological barrier for the initial introduction as it offers users flexible development tools. However, the real challenge is how to process and analyze the physical signals generated from the operating equipment and convert them into practical and useful information to help customers carry out predictive and preventive maintenance. Signal analysis is easier said than done.To accurately analyze diverse physical signals, manufacturers will need help from on-site experts who are familiar with equipment characteristics. The role of on-site experts is similar to that of veteran technicians in the old days. Aside from helping manufacturers raise the efficiency during the process of hardware set-up, on-site experts can also help increase the accuracy of irregularity detection during the process of model building.Citing a case at a factory, Wu found that model building is usually the most time-consuming step during the initial set-up of an equipment monitoring system. This is because the equipment has to maintain non-stop operation until an irregularity occurs so that there are parameters available for analysis and comparison. This can take up to one month or even one year.The process of model building is like a self-learning process for the equipment. By simulating occurrences of irregularity, the equipment monitoring system is trained to make a determination on the cause of irregularity through changes in the physical signals and then proceed to resolve the issue. With a growingly large database of physical signals for analysis and comparison, the accuracy on irregularity determination will significantly increase.Although model building takes time, once you get the ball rolling and successfully get past this stage, all that remains is to follow standard procedures and put the module to work in other pieces of equipment. Subsequent development cycles can then be accelerated. Model building is an essential and intricate step in the set-up of an equipment monitoring system, which cannot be handled by general factory workers.This is why on-site experts can be of help at this stage with data analyses and system configurations based on the structures and characteristics of the mechanical components and the signals generated during operation. The purpose of engaging on-site experts is to ensure manufacturers can quickly build up an accurate model at the beginning of system set-up to protect manufacturers from having to spend more time correcting errors made in the process.Find the best suited monitoring system for individual industryHardware configuration is also an importation consideration in the build-up of an equipment monitoring system in addition to signal analysis and processing. The selection of monitoring devices will also differ based on the manufacturer's requirements, with cost being one deciding factor and business type being the top consideration.Take Advantech's signal capture card for vibration measurement for example. A range of signal capture cards are available to meet the needs of different industries. The main differences are in their ability to simultaneously measure multiple sets of signals as well as their resolution levels. Signal capture cards supporting a higher resolution can detect finer vibration frequencies.Whether simultaneous capture of equipment signals is achievable depends on the sampling frequency and availability of independent channels. Sampling frequency refers to the number of physical signals collected per second. A higher frequency means more samples are captured, which also means more data entries are available for time domain and frequency domain analyses to arrive at accurate judgement. Channel independence allows each channel to be able to collect data independently at the same time.In contrast to sampling frequency, resolution, which defines the fineness of signal changes that can be captured, is another factor that affects monitoring efficacy. Signal capture cards supporting a higher resolution can capture finer changes in the collected physical signals and therefore can detect very subtle vibration changes.Manufacturers in the machine tool industry are generally satisfied if an equipment monitoring system can give them enough time to take preventive actions before failure occurs, thus imposing less stringent requirements on the sampling frequency and resolution. They tend to adopt entry-level low-cost systems to meet their needs. To the semiconductor industry, however, it's a different story.Semiconductor manufacturing lines produce electronic devices with high precision and have to maintain operation around the clock. Unlike machine tools that perform cutting on large-size work pieces, semiconductor manufacturing allows little tolerance for slight deviations so manufacturers in this field of work place great emphasis on instantaneity and synchronization in their selection of monitoring systems. They tend to opt for signal capture cards capable of simultaneously collecting signals so as to prevent misjudgment due to latency.According to Wu, the composition complexity of production equipment is proportional to its level of precision. A piece of sophisticated equipment will generate multiple vibration frequencies during operation. Even minor vibrations from a tiny component could be the cause of equipment irregularity. Therefore, high-precision production equipment such as that used in semiconductor manufacturing requires high sensitivity detection of every subtle vibration signal, which calls for signal capture cards supporting high resolutions.However, a higher end monitoring system does not necessarily offer better signal detection sensitivity. Manufacturers should still choose a monitoring system best suited to their needs. Cost consideration remains the deciding factor when Taiwan industries make their choice on equipment monitoring systems. High-end systems will certainly cost more. Manufacturers will have to weigh the expense against the benefit and decide whether to make such an investment.

In the past, manufacturers generally relied on their experienced technicians to manually check the health of all their equipment during routine maintenance. However, such a practice faces problems.First of all, labor shortage may affect the frequency and efficiency of the routine maintenance. Second, novice technicians may have limited experience working with the multitude of sophisticated equipment, which could compromise inspection reliability. There could even be safety risks in potentially dangerous work environments. Furthermore, with time-based maintenance, manufacturers have difficulty grasping real-time operating conditions of their equipment and they generally are more cost-conscious especially with high-precision and high-cost production equipment.Manufacturers could encounter two situations. One is early replacement of equipment before the end of its useful service life. This unnecessarily increases operating costs if expensive equipment is replaced. The other situation is that manufacturers have no knowledge of equipment irregularity when it starts to happen but the equipment is not yet due for maintenance or replacement. This could result in even more serious consequences.Sheng-Wei Wu, product manager, Industrial IoT Group, Advantech, commented if equipment operates under irregular conditions or outside of tolerance, the production line will have a lower yield. In the case of continuing operation without timely detection of the irregularity, manufacturers face the risks of unexpected downtime and possible damage to peripheral components resulting in extensive loss. The root of these problems that give manufacturers a headache lies in their inability to stay on top of equipment conditions at all times. Traditional time-based maintenance falls short of expectations. By introducing equipment monitoring systems, manufacturers can instead take a proactive approach to use machines rather than manual labor, and scientific methods rather than experience to build up their ability in predictive and preventive maintenance. Equipment health data: physical signalsIn general, equipment monitoring can be divided into a few aspects (or steps). First of all, it is fundamental to have sensing capability on the equipment in order to gain an understanding of its operating condition. This is achievable through today's sensor technologies which can collect physical signals from the equipment in operation. These physical signals which are the equipment's health data serve as the basis for the determination on whether the equipment is maintaining normal operation or the prediction of when a failure might occur.Various types of sensors can be used to detect changes in the equipment's components. Vibration sensors and electric current sensors are more common. As motors are the core components driving equipment operation, the changes in motor electric current are usually the most important factor to equipment monitoring as basis for making judgements. When the equipment is not operating smoothly, more driving power is needed and the electric current will therefore increase. Accordingly, in the case of sudden electric current surge, it could mean the motor is not operating normally.However, when the electric current shows an irregular increase, the equipment may already be on the brink of a failure. Vibration is another physical signal indicating possible equipment malfunction. Vibration detection can enable early discovery of signs of mechanical aging or damage. Wu indicated vibration monitoring allows the detection of even a slight displacement of components in the equipment so that manufacturers have ample time to deal with aging or malfunctioning parts before potential failure actually happens.In addition to the common electric current and vibration signals, there are other sensing mechanisms available for detecting changes in equipment components. For example, thermal sensing can be used to measure heat generated from equipment movement, such as friction caused by machining. Detections of noise, pressure and spindle speed are other sensing technologies useful to equipment monitoring. Different types of equipment may have different failure modes, which may manifest themselves with various physical indications. Wei-Han Wu, engineering marketing manager, National Instrument, drew an analogy between equipment monitoring and medical diagnosis. A doctor considers various measurements such as heart rate, blood pressure, pulse and breathing and then arrives at a diagnostic decision. Equipment maintenance experts also conduct health evaluations on machinery based on various physical information.On-site experts are indispensableAfter the various types of sensors gather physical signals from the equipment in operation, the next step will be for the signal capture module of the equipment monitoring system to collect and report the signals back to the system host. Further analyses and processing by inspection software can determine whether equipment components are working normally. For a large-scale manufacturer operating on a multi-site or multi-national basis, remote monitoring is also feasible with cloud-based signal communication.Before introducing an equipment monitoring system, the manufacturer should first evaluate its factory environment for considerations such as hardware wiring. As the equipment chassis may have limited space to accommodate monitoring devices, along with the wiring, space arrangement may become challenging. Maintenance and management difficulty could also be an issue.The first and foremost benefit of equipment monitoring is that manufacturers can stay aware of equipment's operating conditions and life cycle through real-time physical data gathered during operation. General IT employees at factories are not trained to deal with the sophisticated and specialized data collected from the equipment. Therefore, on-site experts (or equipment service experts) skilled with various types of machinery play an important part in the buildup of equipment monitoring systems.Wu said on-site experts with comprehensive knowledge on structures and operations of various machinery can further help factory management raise efficiency when putting equipment monitoring systems in place and make accurate decisions on the basis of data analyses.For example, on-site experts with their know-how on the machinery have better knowledge than anyone else about where the equipment may go wrong. They can therefore help factory management put sensors at the right places to ensure they accurately and effectively collect signals during equipment operation.In addition, on-site experts also serve as a bridge between factory equipment and back-end IT professionals to help them convert the collected raw data into useful information facilitating the determination of normal operation.

The manufacturing industry has always been fundamental to Taiwan's economic prosperity. However, the Made in China 2025 initiative and manufacturers being urged to return to the US have begun to worry the Taiwan manufacturing industry, which could be replaced by its counterparts in China and South Asia in five years if it fails to upgrade. Industry 4.0 smart manufacturing promises to enhance their core competitiveness.Advancing technological developments such as artificial intelligence (AI), big data analytics and online/offline integration all fall within the scope of Industry 4.0 and are being put to use in wide-ranging applications. Among them, equipment monitoring technology can produce immediate results for the Taiwan manufacturing industry. Production equipment is instrumental to manufacturing. Much like a craftsman who can to do the job only with a good tool, manufacturers need their production equipment to maintain normal operation to keep their production lines up and running smoothly.Health checks for production equipment used to be done through routine services, during which machines are halted for inspection, maintenance and replacement. Despite such routine services, manufacturers are still uncertain about the reliability of their production equipment as there is still no guarantee that the equipment will be fail-proof after the services.Production equipment does not talk to operators to tell them when it is operating normally and when it is going to fail, so it is usually a disaster if downtime occurs due to equipment failure. Engineers will have to rush to the site to conduct checks and emergency repair. It is a race against time because equipment shutdown also means production halt with far-reaching consequences.As such, the question is whether there is a preemptive way to solve the problem before a failure occurs and results in production line shutdown. This is exactly the job for equipment monitoring. Different from traditional time-based maintenance, equipment monitoring is all about using real-time status checks to constantly stay aware of equipment conditions during operation. Much like a doctor using diagnostic instruments to check a patient's physiological data, equipment monitoring makes use of various types of sensors to gather equipment health data such as electric current, vibration and noise to determine if it is functioning normally.Big data analytics is then used to help operators predict the equipment's service life based on its actual operating status. Factory management can thereby make informed decisions according to actual findings collected during equipment inspections. Furthermore, Big data analytics can assist in the prediction of potential failure risks so that operators can make a preemptive response by adjusting production lines before the failure actually happens to prevent the production from being halted due to unexpected failure and thus enhance production stability. As a matter of fact, equipment monitoring can also be valuable to people's everyday life. Take vehicle belts for example. They will show signs of wearing or aging over time. Unlike other automobile parts, they must be replaced before breakdown. A broken belt could pose serious threat to road safety so automakers generally request that car owners regularly check and replace vehicle belts. A vehicle belt that does not show cracks does not necessarily mean there is no problem. Whether the belt needs to be replaced is to be determined by the mechanic based on his experience rather than scientific evidence. But this is hardly a safe practice.In fact, how fast a belt wears out depends largely on how the driver uses the car. However, it may be difficult to track each driver's use habit so whether to replace a belt can rely on either the mechanic's experience or routine maintenance. From another perspective, if the belt is only mildly worn, routine check and replacement incurs extra costs to the car owner.If there is a way to objectively assess the wear and health conditions of the belt, rather than relying on individual mechanics' experience, the car owner may be able to take the car to maintenance at more precise timings. The concept applies to manufacturing equipment monitoring as well. However, to a manufacturer, when production equipment failure occurs, it is not as easy as changing a belt.For instance, should any equipment failure or unexpected emergency happen on a semiconductor manufacturing line that is supposed to operate around the clock, resulting in production shutdown, it could result in a multi-million-dollar loss. Aside from the loss of scrapped materials, foundries sustain even more pressure for failing to meet delivery schedule and stand to lose customers' trust.Equipment maintenance is necessary. The worst case scenario is when unexpected failure occurs and emergency repair is needed. As this is unpredictable, it is difficult to estimate the downtime required for repair and the potential loss. Time-based maintenance is a safer approach but inevitably more costly.Predictive and preventive maintenance made possible through equipment monitoring allows manufacturers to stay aware of real-time equipment operating conditions and foresee potential equipment failure. This is no doubt the most practical approach meeting the needs by smart manufacturing toady.

Along with mature development of ECG (electrocardiography) and PPG (photoplethysmograph) sensing technologies, ICT firms are keen on developing monitoring devices for use in mobile medical care, which is expected to see demand take off, according to Digitimes Research.Mobile medical care will be initially applied to diabetes and cardiac arrhythmia, Digitimes Research said.Global market value for mobile medical care will increase to US$189 billion in 2025 at a compound annual growth rate of 32.3% during 2017-2025, research organizations have forecast.US Food and Drug Administration in the third quarter of 2017 launched Digital Health Software Pre-certification Pilot Program to facilitate development of devices for mobile medical care.Growing demand for mobile medical care will drive demand for bio-sensor chips, accelerator chips, MCUs (micro-controller units) and wireless communication (such as Bluetooth) modules.

Artificial intelligence (AI) is deemed as a critical technology needed to boost LPWAN (low power wide area network) application values. Global tech giants including Amazon, Microsoft, Google, Baidu and other software service providers are aggressively developing the AIaaS (artificial intelligence as a service ) business model, and telecom operators are seeking to integrate AIaaS service s and their massive MTC (machine type communication) data generated by LPWANs to explore immense potential business opportunities in the new era of AI, according to Digitimes Research.The NB-IoT and eMTC standards released by 3GPP for cellular LPWAN applications in mid-2016 involve low investment costs and high network coverage for current 4G LTE operators, but only 3.6% of global LTE network operators had launched such cellular LPWAN commercial services as of the end of October 2017.The reluctance of the majority of telecom operators to launch LPWAN services is partly caused by their little knowledge about how to boost LPWAN application values, Digitimes Research believes. The application values of LPWAN evolve in three stages: data collection, data analysis and forecasting. As far as long-term industrial values are concerned, LPWAN should serve as important foundation for AI machine learning and forecasting capability while AI will impact telecom operators in three major aspects: consumer services, big data analysis and network architecture.

The in-vehicle networks currently used in automobiles are based on a combination of several different data networking protocols, some of which have been in place for decades. There is the controller area network (CAN), which takes care of the powertrain and related functions; the local interconnect network (LIN), which is predominantly used for passenger/driver comfort purposes that are not time sensitive (such as climate control, ambient lighting, seat adjustment, etc.); the media oriented system transport (MOST), developed for infotainment; and FlexRay for anti-lock braking (ABS), electronic power steering (EPS) and vehicle stability functions.As a result of using different protocols, gateways are needed to transfer data within the infrastructure. The resulting complexity is costly for car manufacturers. It also affects vehicle fuel economy, since the wire harnessing needed for each respective network adds extra weight to the vehicle. The wire harness represents the third heaviest element of the vehicle (after the engine and chassis) and the third most expensive, too. Furthermore, these gateways have latency issues, something that will impact safety-critical applications where rapid response is required.The number of electronic control units (ECUs) incorporated into cars is continuously increasing, with luxury models now often having 150 or more ECUs, and even standard models are now approaching 80-90 ECUs. At the same time, data intensive applications are emerging to support advanced driver assistance system (ADAS) implementation, as we move toward greater levels of vehicle autonomy. All this is causing a significant ramp in data rates and overall bandwidth, with the increasing deployment of HD cameras and LiDAR technology on the horizon.As a consequence, the entire approach in which in-vehicle networking is deployed needs to fundamentally change, first in terms of the topology used and, second, with regard to the underlying technology on which it relies.Currently, the networking infrastructure found inside a car is a domain-based architecture. There are different domains for each key function - one for body control, one for infotainment, one for telematics, one for powertrain, and so on. Often these domains employ a mix of different network protocols (eg, with CAN, LIN and others being involved).As network complexity increases, it is now becoming clear to automotive engineers that this domain-based approach is becoming less and less efficient. Consequently, in the coming years, there will need to be a migration away from the current domain-based architecture to a zonal one.A zonal arrangement means data from different traditional domains is connected to the same ECU, based on the location (zone) of that ECU in the vehicle. This arrangement will greatly reduce the wire harnessing required, thereby lowering weight and cost - which in turn will translate into better fuel efficiency. Ethernet technology will be pivotal in moving to zonal-based, in-vehicle networks.In addition to the high data rates that Ethernet technology can support, Ethernet adheres to the universally-recognized OSI communication model. Ethernet is a stable, long-established and well-understood technology that has already seen widespread deployment in the data communication and industrial automation sectors. Unlike other in-vehicle networking protocols, Ethernet has a well-defined development roadmap that is targeting additional speed grades, whereas protocols - like CAN, LIN and others - are already reaching a stage where applications are starting to exceed their capabilities, with no clear upgrade path to alleviate the problem.Future expectations are that Ethernet will form the foundation upon which all data transfer around the car will occur, providing a common protocol stack that reduces the need for gateways between different protocols (along with the hardware costs and the accompanying software overhead). The result will be a single homogeneous network throughout the vehicle in which all the protocols and data formats are consistent. It will mean that the in-vehicle network will be scalable, allowing functions that require higher speeds (10G for example) and ultra-low latency to be attended to, while also addressing the needs of lower speed functions. Ethernet PHYs will be selected according to the particular application and bandwidth demands - whether it is a 1Gbps device for transporting imaging sensing data, or one for 10Mbps operation, as required for the new class of low data rate sensors that will be used in autonomous driving.Each Ethernet switch in a zonal architecture will be able to carry data for all the different domain activities. All the different data domains would be connected to local switches and the Ethernet backbone would then aggregate the data, resulting in a more effective use of the available resources and allowing different speeds to be supported, as required, while using the same core protocols. This homogenous network will provide "any data, anywhere" in the car, supporting new applications through combining data from different domains available through the network.Marvell is leading the way when it comes to the progression of Ethernet-based, in-vehicle networking and zonal architectures by launching, back in the summer of 2017, the AEC-Q100-compliant 88Q5050 secure Gigabit Ethernet switch for use in automobiles. This device not only deals with OSI Layers 1-2 (the physical layer and data layer) functions associated with standard Ethernet implementations, it also has functions located at OSI Layers 3,4 and beyond (the network layer, transport layer and higher), such as deep packet inspection (DPI). This, in combination with Trusted Boot functionality, provides automotive network architects with key features vital in ensuring network security.(Christopher Mash is senior director of Automotive Applications & Architecture at Marvell)