Challenges of IoT
At Enterprise Scale
Fybr’s tightly integrated technology stack has been proven to tackle these challenges successfully while enabling a fast path to ROI for IoT projects.
Scale of data and integration complexity
Connected devices create volumes of data, which are unprecedented in traditional enterprise systems.
Cisco estimates6 that the total volume of data generated by the Internet of Things will reach 600 Zettabytes per year by 2020, which is 275 times higher than projected traffic going from data centers to end users/devices (2.2 Zettabytes) and 39 times greater than total projected data center traffic (15.3 Zettabytes).
Understanding data of this size in the context of enterprise transactions and transforming it into actionable information mandates a data storage, integration, and analysis infrastructure that is designed from the ground up to handle these volumes, cut across silos, and produce results within time spans that deliver an advantage
to the business.
Fast moving data and decision latency
In addition to being large-scale, device data moves rapidly. The speed at which 10,000 human users can create data is orders of magnitude slower than the rate at which 10,000 automated devices can produce messages. Enterprise systems that were traditionally designed to handle thousands of simultaneous user transactions may now have to respond to millions of device events.
To remain efficient, systems must be capable of comprehending data moving at much higher velocities while business processes need to be redesigned for faster decisions and quicker reaction time.
Real world conditions and sensing accuracy
In most applications, the only way to accurately sense a physical event or condition is to use learning algorithms that can be individually tuned to the surroundings of a particular device in a network that may contain thousands of similar programmable devices. To add further complication, environmental conditions may change over time, requiring constant adjustments to maintain accuracy.
Sustained accuracy and reliability is only achievable by enabling the sensing algorithm to learn and evolve with environmental conditions and business context.
Location specific power constraints
Many urban, industrial, and agricultural IoT applications envision devices in areas where providing power can be a challenge. Additionally, power consumption can be a crucial element in the successful implementation of a remote IoT system. Sensors and actuators installed in underground or remote pipelines, inside road beds, inside containers, etc. are all examples of devices that must depend on either battery power or must somehow harvest enough energy from their environment.
In either case, these devices must be designed to be extremely efficient in their use of power. Many of these applications can only be served by low data-rate, low-power edge devices that have been specially engineered to operate in such environments, as communications are often the biggest single factor in an edge device’s power consumption.
Network reliability in urban and industrial environments
Similar to power constraints, wired network connectivity is not feasible in many scenarios, requiring IoT solutions to rely on wireless networks that have an extremely low-power footprint. Poorly designed protocols, bloated messages or inefficient encodings can all prove to be difficult engineering hurdles.
Moreover, industrial and urban environments are often affected by radio/electromagnetic noise or physical obstructions. Industrial IoT networks must be designed to adapt to such noisy conditions while providing redundant communication paths without compromising system power efficiency.
Often, the information an IoT application requires cannot be confirmed by just one device. For instance, to detect a blockage in a sewage system, information from multiple sensors must be read collectively.
IoT systems must enable this type of concurrent collaboration between edge devices. Sometimes this communication is direct, over physical channels like radio, at other times it may require coordination at a central server. Direct communication may be instantaneous but can come at a significant power cost. Centralized collaboration may help reduce power consumption and enable cooperation between devices, but may increase latency.
Direct vs. central collaboration can present a range of trade-offs between power consumption, latency, and cost. A modern solution must empower applications that can prioritize any of these factors based on the appropriate business need.
Provisioning and maintaining thousands of distributed devices
- Building, installing, and operating a network of connected things with tens of thousand of geographically distributed devices can present unique logistical challenges. Ensuring quality of service requires a system that can precisely track the entire life-cycle of each device from manufacturing to recycling.
- Before installation, such a system must track firmware versions loaded on each device, record unique identifiers, generate encryption keys, and record results of manufacturing quality checks.
- During installation, information such as when a device was installed, which team installed it, the exact install location, configuration, and activation time must all be collected.
- In operation, any significant event, maintenance activity, configuration history, firmware version history, suspicious event, etc. must all be tracked to enable future diagnosis and audits.
Flexibility, adaptability and risk in updates
Adopting the Internet of Things is a strategic investment for any enterprise. It is not enough to only consider short-term needs. IoT deployments must adapt to changing requirements while supporting future connected solutions that have yet to be created.
Initial IoT projects will often be experimental; therefore systems must enable simple and safe iterations, allowing teams to improve business logic and discover the best way to integrate physical world data into their processes.
IoT systems must also be flexible enough to easily integrate new sources of internal or external data that may offer additional business context.
To facilitate these evolutions, over-the-air firmware updates for remotely deployed devices is a critical, yet highly risky operation, traditionally leaving a system open to device failure and security vulnerabilities. A poor implementation may allow for bad updates that may, in turn, render a device inoperable or unable to communicate, or even render an entire IoT device network unusable—requiring costly repair efforts and potentially lengthy downtime.
To be successful, IoT solutions must enable a safe method of efficiently updating remote devices without leaving the business critically vulnerable.
A 2016 paper by the US Department of Homeland Security (DHS)7, on IoT device security, states that “Many of the vulnerabilities in IoT could be mitigated through recognized security best practices, but too many products today do not incorporate even basic security measures. There are many contributing factors to this security shortfall. One is that it can be unclear who is responsible for security decisions in a world in which one company may design a device, another supplies component software, another operates the network in which the device is embedded,
and another deploys the device.”
While it is essential to enforce established enterprise security practices with strict authentication, careful access management, granular network policies, etc., it is also prudent
to avoid ad-hoc piecemeal solutions where there is no well-defined owner of security for the entire loosely integrated stack of technologies.
IoT solutions also present many new security challenges.
Mainstream authentication and encryption protocols are far too computationally expensive for most low-power wireless devices. Asymmetric algorithms are often impractical, while symmetric encryption presents complex key distribution and key management challenges.
Further, many IoT devices installed in urban and industrial scenarios are not protected by physical boundaries, leaving them vulnerable to tampering. This makes it hard for such edge devices to guarantee safe storage of keys. These challenges make robust key life-cycle management crucial to ensuring unique keys for every device. Additionally, new approaches to detect tampering and take remedial actions are necessary for sustaining a secure and reliable system.
Technology landscape fragmentation
There are many IoT solutions available on the market; however, most vendors provide only a subset of the components required to implement an integrated end-to-end solution.
Some vendors provide just the edge devices, others provide just the low-power wireless network, while many others sell “IoT platforms” which, in most cases, are just the centralized software-only server component with management tools.
Often, system integrators cobble together edge devices, wireless networks, and server-only IoT platforms to build a solution. As noted above, such piecemeal solutions often suffer from security issues due to the tremendous challenges of ensuring tight controls and carefully managed key life-cycles when so many vendors are involved.
Additionally, these cobbled solutions can also suffer from poor power efficiency, network reliability, and sensor accuracy. Piecing together a reliable network of devices requires extensive testing in real-world environments and several iterative improvements, taking inordinate amounts of time to achieve a desirable quality of service,
at scale. If it takes too long, this may cause
projects to fail.
Skills and experience gap in the industry
Most businesses now have teams experimenting with IoT, yet very few of those teams have experience with large-scale, geographically distributed IoT deployments. This can result in unexpected costs, risks and delays in real enterprise scale projects.
Scale of data and integration complexity