A Guide to Processors for IoT and Wearables, Second Edition
|IoT・ウェアラブル向けプロセッサーのガイド A Guide to Processors for IoT and Wearables, Second Edition|
|出版日: 2016年04月27日||ページ情報: 英文||
モノのインターネット (IoT) 市場はその多様性とIoTが正確には何を意味するのかという明確性に欠けていることから、測定が難しくなっています。多くの予測専門家は、2020年までに活動するIoTデバイスが250億、もしくは500億、さらには750億に上ると推計していますが、これらの曖昧な予測には大抵スマートフォン、PCおよびその他既存のクライアントデバイスが含まれています。当レポートでは個別の漸進的な機会を明確にするため、IoT市場を区分別に調査しています。また、求められる価格帯・技術的性能も明らかにしています。
The Internet of Things (IoT) is a buzzword that encompasses a wide range of end products, ranging from industrial control to home automation, from smart utility meters to networked refrigerators, from connected cars to smart watches. Almost any embedded system can be turned into an IoT device by simply adding a wired or wireless Internet connection. This simplicity enables a rapid pace of innovation - some good and some not.
The IoT market is difficult to measure because of its diversity and the lack of clarity as to what exactly IoT means. Many forecasters expect 25 billion or 50 billion or 75 billion IoT devices will be active in 2020, but these vague estimates often include smartphones, PCs, and other existing client devices. We instead examine the IoT market segment by segment to identify specific incremental opportunities. This approach also exposes the price points and technical capabilities that will be required.
Today, many IoT products use standard embedded processors or microcontrollers (MCUs), but the market is still quite nascent. Although industrial devices, such as smart meters, may cost more than $100, consumer devices will require lower prices to enable the expected growth. As the bill-of-materials (BOM) cost, or incremental BOM in the case of an appliance or car, falls below $10, low-cost processors and connectivity chips are required. In many cases, the processor and wireless chips will come from the same supplier or will be integrated onto the same chip. This approach simplifies the system design and reduces cost.
Many chip vendors are already developing or delivering products that combine a processor and a wireless connection. Depending on the target applications, the type of connectivity could be Wi-Fi, Bluetooth, Zigbee, or proprietary sub-1GHz protocols. "A Guide to Processors for IoT and Wearables" focuses on this emerging product category of integrated IoT processors.
Unlike typical market research, this report provides technology analysis and head-to-head product comparisons. Which chips will win designs and why? How will these vendors be positioned as the Internet of Things continues to grow? Only The Linley Group's unique technology analysis can provide this forward-looking view.
"A Guide to Processors for IoT and Wearables" begins with an extensive overview of this complex market, breaking it down into several segments such as industrial, smart meters, home automation, connected appliances, and wearable devices. For each segment, we discuss the processor and connectivity requirements as well as cost and power. We also provide a forecast for each segment, indicating how the IoT is likely to grow.
Following the introductory chapters, the report delivers detailed coverage of integrated IoT processors from Broadcom, Freescale, Marvell, MediaTek, Realtek, Texas Instruments, and Toshiba. It also covers IoT-focused processors from Atmel, Ineda, Intel, NXP, Samsung, Silicon Labs, and STMicroelectronics. For each of these vendors, we discuss the company's strategy for IoT, its relevant products and technologies, and expected future directions. The report includes technical specifications for the relevant announced products and discusses how well they fit their target applications.
The report concludes with a comparison of these products, broken down by application, to highlight which products and vendors are best positioned to succeed in IoT. Finally, we offer our outlook for the leading vendors and for the overall market.
As the leading vendor of technology analysis for processors, The Linley Group has the expertise to deliver a comprehensive look at the full range of chips designed for a broad range of IoT applications. Principal analyst Linley Gwennap and Senior Analysts Mike Demler and Loyd Case use their experience to deliver the deep technical analysis and strategic information you need to make informed business decisions.
Whether you are looking for the right chip for your IoT application or seeking to partner with or invest in a chip vendor, this report will cut your research time and save you money. Make the intelligent decision, order "A Guide to Processors for IoT and Wearables" today.
Engineers designing IoT or wearable chips or systems
"A Guide to Processors for IoT and Wearables" has been updated to incorporate new announcements made since the publication of the previous edition.
Here are some of the many changes you will find:
The "Internet of Things" (IoT) is one of the most overused terms in the semiconductor industry, yet its definition remains unclear. Some people use the IoT label for any Internet-connected device, including set-top boxes, smartphones, and PCs. Rather than applying this label to large established markets, we prefer to focus on new markets, allowing us to gauge IoT's true incremental value. A new class of processors is emerging to meet the unique requirements of these markets.
Initial forecasts of 50 billion IoT devices by 2020 were wildly optimistic, even though they included all Internet devices. Our forecast - which excludes established markets - predicts shipments of 1.9 billion IoT and wearable devices in 2020. By comparison, we expect 2.2 billion smartphones to ship that year. Our forecast shows a total installed base of about 6 billion devices in 2020; adding 10 billion to cover established markets brings the total to 16 billion Internet-connected devices that year.
Our forecast assumes consumer devices will become the biggest contributor to IoT volume, as consumers far outnumber factories and farms. We limit the target market for consumer IoT to the 600 million households worldwide that will have Wi-Fi and broadband connectivity in 2020, not the seven billion individuals on the planet. Because the value proposition for IoT products (except smart meters) remains unproven, forecasts for the number of IoT devices per household, factory, or farm are quite speculative. Our model shows an average of nine IoT devices per target household in 2020.
To reach this deployment rate, low cost is essential. Initial prices include $349 for a Galaxy Gear smartwatch, $250 for the Nest thermostat, $278 for the Goji smart lock, and $79 for a Lifx light bulb. These prices have already dropped; we expect their price to fall below $100 and in many cases below $50. Some reduction will come from economies of scale and lower margins, but to achieve these price cuts, the bill-of-materials (BOM) cost must also fall.
As in other high-volume markets, application-specific semiconductor products (ASSPs) will emerge, reducing BOM cost through integration. Because most IoT devices have a wireless connection, integrating the processor and the radio is becoming common. IoT processors often integrate memory (RAM and ROM) as well. Analog interfaces (including analog/digital converters) can connect the processor to sensors, motors, and actuators. This level of integration also reduces power consumption, which is critical in wearables and other battery-operated devices, such as remote sensors.
Because IoT covers many different applications, processing and wireless requirements vary. Many IoT devices perform little processing, merely uploading their data to a cloud service and executing simple commands. Others, such as security cameras, may analyze the data locally. Devices with a user interface require more processing. Most consumer IoT devices use Wi-Fi, but to reduce cost and power, others use IEEE 802.15.4 or similar protocols. Wearables typically communicate with a nearby smart-phone using Bluetooth. Vehicles and some industrial devices rely on cellular communications.
Many chip vendors have already jumped into the IoT market. Broadcom, Marvell, MediaTek, and Realtek have added a programmable CPU to their existing wireless chips to create integrated IoT processors. Many of these chips include a proven Wi-Fi radio and are well suited to consumer IoT devices. Marvell is the only vendor supporting the full range of IoT protocols: cellular, Wi-Fi, Bluetooth, and ZigBee (802.15.4).
Atmel, NXP, Silicon Labs, STMicroelectronics, and Texas Instruments have instead added wireless to their existing microcontrollers. These processors have limited CPU horsepower, but they often integrate memory and analog interfaces to reduce system cost. Most of them support simple protocols such as 802.15.4 and Bluetooth, which also minimizes power. TI is the only vendor in this group that has integrated Wi-Fi, based on its former mobile-connectivity products.
Many wearables shipping today (including the Apple Watch) use repurposed smartphone processors. MediaTek has gone the extra mile and designed a highly integrated processor for smartwatches. Qualcomm rolled out its first processor branded for wearables. Samsung offers the Artik 1 module, which uses a custom processor from startup Ineda.
Intel generates large "IoT" revenues from its embedded x86 processors, but it has yet to integrate x86 and wireless (other than cellular) on the same chip. The company's Curie module combines its Quark SE processor with Bluetooth, memory, and analog components. This multichip approach is power hungry and costly, but it could eventually lead to an integrated solution.
Each of these initial processor offerings has significant shortcomings that prevent it from addressing the entire IoT market. As long as IoT demand remains unproven, chip vendors will be hesitant to invest large sums in ASSP development. Once a few IoT devices achieve high volumes, some vendors will improve their products to meet the needs of those applications. But the IoT market is broad enough to accommodate a large number of suppliers. This report explains what IoT and wearable devices each vendor can address today, as well as what technologies each can use to develop future ASSPs for these applications.