The Internet of Things (IoT) hype is now getting real. This, in turn, is creating the opportunity to move from off-the-shelf chip designs to custom silicon. The key to creating cost-effective, custom silicon for the IoT will be the platform approach.
It is fair to say that IoT is now living up to the hype. There has been a significant uptick in activity over the past year with the IoT ecosystem, the end customers, the hardware and software vendors, the system integrators and the startup community. Yes, IoT implementations are definitely happening, although not at the same rate as first expected, and certainly not the 50 billion devices or a trillion sensors by 2020. Nevertheless, it is an encouraging sign for ASIC design companies as they become an important and differentiating cog of the IoT supply chain.
Historically the industry has been churning out custom silicon for the cloud side of IoT for years, specifically in the networking, telecommunication, storage and computing arenas. However, devices on the edge of the IoT network have, so far, been designed using stock components rather than custom silicon. Using a platform approach to custom silicon design can substantially enhance functionality and offer greater design flexibility.
An IoT edge device typically performs three functions: sensing/actuating, processing and communication. Depending on various factors, like cost, schedule and application, a custom silicon implementation in these IoT edge devices could be a low-end/low-cost, mid-level or highly integrated solution. Below is a representation of typical ASIC configurations of each type. In all of these cases, ultra-low power is a default requirement.
Example ASIC Configurations
At the start of the IoT era, most designers were satisfied using standard ICs/ASSPs to make edge products. Custom silicon for these devices did not cut it, either because of the NRE cost (although unit price can be way cheaper) or the product schedule would kill an ASIC plan.
Recently, however, designers across vertical markets like healthcare, industrial and smart utilities, want custom silicon to differentiate their products and future-proof them as much as possible. In many cases, these designers are working on their second- or third-generation product lines and they have hit sufficient volumes to justify the NRE investment required for ASICs. In other cases, customers feel that ASSPs don’t offer the differentiation they desire, or they have an IP, or secret sauce, that needs to go in hardware form.
Nevertheless, the schedule issue still remains. This is where IoT ASIC platforms become very relevant. Done right, platforms can speed custom design while retaining the ability to differentiate. An IoT edge device, based on an ASIC platform, should cater to the following goals:
Use FPGA setup to provide:
- A ready-to-use development setup to meet the needs of both hardware and software developers.
- A vehicle to demonstrate a proof of concept including, in some cases, a gateway and cloud back-end.
- A scalable evaluation setup for trying out different HW-SW partitioning or a custom HW/SW IP.
- Provide an ASIC environment that allows power benchmarking and evaluation of different power management schemes.
Creating IoT ASIC platforms requires thinking like a system company, or even like a startup, and requires the consideration of end use-cases in the various IoT vertical markets. This also requires creating an end-to-end system with edge devices, and in some cases, including a gateway and a cloud back-end. Some scenarios that can utilize IoT edge device ASIC platforms include:
- Ambient environment sensing in an industrial environment with communication protocols like LoRa, W-HART and Zigbee. The system view of this use case is shown below.
- Motion sensing using 9-axis sensors (Accelerometer, Gyrometer and Magnetometer) and BLE communication.
- Pulse oximetry with BLE communication.
Use-case: Ambient Environment Sensing & Communication in an Industrial IoT Setup
Creation of IoT ASIC platforms for the industrial use case, for example, requires a two-phase approach. The first phase is to create an end-to-end IoT setup with gateway and cloud back-end, using standard ICs/ASSPs to design the edge device hardware. This approach enables creating demo scenarios quickly, putting hardware and software pieces at the edge, gateway, and cloud together as shown in the figure below.
Phase 1 IoT System Demonstration Setup
The second phase calls for replacing the ASSP-based edge device hardware with FPGA-based platforms, creating reference SoC designs, along with complete software stacks and peripheral devices, to address different use-case scenarios. The figure below shows an FPGA platform for use in developing reference designs.
In parallel, to address the low power requirement, the platform needs an ASIC RTL design database (from the same SoC specification used for an FPGA platform). The idea is to create a power benchmarking setup to evaluate different power management schemes, such as power domains, always-on subsystem, custom ultra-low power memory, cells and more.
FPGA Platform for IoT Edge ASIC Designs
Implementing reference designs on the FPGA platform requires an ecosystem around hardware IP vendors, including partnerships, such as the one shown below. The software side includes leveraging from the open-source community.
IoT Edge Device ASIC Ecosystem
Such platforms will help IoT edge device design teams to quickly create differentiated designs targeting a custom silicon implementation, while fostering knowledge re-use in future designs. For an ASIC design house, such as Open Silicon, a platform has additional benefits. It allows for demonstrations at conferences and trade shows, and allows sales and technical solution teams to go out and have meaningful conversations with potential customers. The platform also serves as a learning tool for engineering teams to quickly ramp up on the new technologies related to security and communication protocols.
Pradeep Sukumaran, is a Senior Solutions Architect at Open-Silicon. He has over 15 years of experience in the embedded domain and works on the pre-sales team interacting with customers on ASIC based solutions. He is also responsible for the IoT program at Open-Silicon. He can be reached at email@example.com.