Five years ago, the network edge was well-defined and fairly narrow, and the types of devices connected to it consisted primarily of personal computers, either laptop PCs or desktops, as well as terminal platforms. Now the network edge has become much deeper and varied, with the addition of a broad and rapidly expanding assortment of new connectivity-enabled “things,” commonly referred to as the Internet of Things (IoT).
The majority of applications and services that will monitor, manage and control the rapidly growing fog layer of IoT devices will be cloud based. An IoT gateway enables network connectivity between IoT devices operating within the fog layer and the cloud. To be effective, an IoT gateway must be able to support the local area connectivity medium and protocols used by the various IoT devices, as well as the connectivity and protocols required to attach to the wide area network, and in turn, to Internet-based cloud applications. Let’s drill down further on some of these key design considerations for an IoT gateway:
Depending on the type of IoT device, the fog layer connectivity used to enable (M2M) network connectivity may take the form of a wireless physical layer, such as ZigBee, Sub-gigaHertz, Z-Wave, low power Bluetooth, low power Wi-Fi or Near-Field Communication. For short range, tethered devices, they may use a wired physical layer such as Home Plug Green PHY. In order to accommodate the wide range of connectivity solutions being used by fog layer devices, an IoT gateway must provide enough attachment ports, of the proper type, to support the modems required to connect to the fog layer devices. The attachment ports commonly used by fog layer modems are typically low data rate, serial based input/output (IO) ports, such as UART, SPI or I2C.
In addition to support for the various physical layer connectivity used by fog layer devices, the IoT gateway must be able to support the protocol used by each type of IoT device. These protocols can include Home Automation, Smart Energy, 802.11n, 6LoPAN and several others. Each type of protocol requires a certain amount of processing workload in order to manage the protocol stack associated with each.
The need to protect data transmitted across a network is well recognized, and there are several mechanisms available today for doing this in access points, as well as laptops, using Wi-Fi or Ethernet security protocols. In addition to the capability of encrypting data, protection against the introduction of rouge software that can compromise security of data, or allow unauthorized capture of data prior to encryption is necessary. These features are essential in IoT applications, where many edge of network devices and sensors will be capturing and transmitting user-specific data between nodes. Since this data can be directly related or linked to an individual user, it is essential that the data be encrypted. This is increasingly being regulated and monitored by legislation, which extends to the specification of encryption standards and protocols to be used. The inevitable result will be a requirement that communication processors used in M2M or IoT applications must have the capability of performing cryptographic operations, like hashing, signing and encrypting data, as well as a secure key storage unit, in order to meet regulatory requirements. Providing this type of security requires a full array of data protection mechanisms, including secure boot, support for a trusted architecture, such as ARMÂ® TrustZoneÂ®, and the ability to detect physical tampering of a device enclosure or board.
The ability to support continuous operation in a communication platform requires protection against Alpha particle effects on memory arrays. As silicon process technology has advanced, enabling smaller and smaller design nodes, the probability of errors due to Alpha particle effects within memory arrays has gone up significantly. This is achieved by incorporating error detection and correction (ECC) technology on all memories, including Layer 1 and 2 caches, as well as SRAM and external DDR memory for maximum reliability, as well as watchdog timers.
Physical connectivity, protocol support, security and reliability are all key design considerations of an IoT gateway. As the number of fog layer IoT devices continues to proliferate the interoperability of these different considerations across multiples devices and gateways will quickly become another key factor for consumers looking to connect a myriad of fog devices to the cloud.
Sources for market size: IMT and InStat reports, Q4-2013 and John Chambers, Cisco CEO.