Providing a basis for IoT topology.
An analysis of IoT network component interplay.
Continuing and emergent concerns with the IoT design sphere.
IoT networking components include speed, safety, and precision in processing
The Internet of Things (IoT) can be described functionally as the fusion of three layers – perceptivity, networking, and application – and each holds a crucial role. Sensors detect and sample actionable information, networks foster communication between devices and process large data stores, and applications provide an interface for interaction between users and systems. The network layer in particular invokes the defining aspect of IoT: connectivity between devices that were formally isolated and incapable of on-the-fly, automated adjustments. This communication results from components of network protocol design that are developed to best suit the strengths and requirements of the application.
IoT networking components need to account for the large amount of data that needs to be sent in short windows with high levels of accuracy. To achieve this, a network requires interworking elements that can perform robustly while anticipating present and future concerns in the privacy and security sector.
What Shapes the Needs of IoT Networks?
IoT is already well integrated into many levels of user appliances and industrial equipment, but greater tendencies towards efficiency in both markets are set to push adoption levels even higher. Central to the idea of IoT is providing communication avenues for device-device communication, which can be further extrapolated into device-user notifications. Because IoT covers a broad range of products and industries, it’s impossible to craft a one-size-fits-all network solution; instead, systems engineers look to balance the optimum performance criteria.
Of outsized importance in IoT is the layout, or topology, of the network. Networks can be comprised of simple roads where nodes are linked end-to-end wherein data can only flow in two directions, or more complicated arrangements with several pathways that bolster redundancy and least-step routing while requiring greater computational ability and resource investment. The topology should echo the reliability needs of the particular implementation, with systems operating with a larger emphasis on speed or criticality of signal transceiving building out enmeshment in their design. The exact type of network, or any modifications therein, will be heavily dependent on a few core performance metrics.
Additional Granularity: Digging Into IoT Networking Components
Specifying to a more base level of IoT networks, design converges on three interlinked factors.
Certain IoT systems (for example, computer vision) are contingent on larger files and more frequent updates for real-time analysis. Processing this data requires greater edge-computing performance, whether local/onboard or server-side. Supporting this IoT configuration also demands high throughput and capture rates of sensors, another impediment to cheap and low-power electronics, which contrasts common IoT systems where cost and low-power operation are often favored due to scale. Intelligent neural network models can be adapted to IoT devices, assigning resource-intensive tasks to the more sophisticated memory and processors while allowing some level of execution at the device level. Encoders can also assist overall performance by systematically reducing data sizes before transfer and decoding with the dedicated processing hardware.
Unfortunately, neural networks tend to suffer in identification when performed over many layers, and care needs to be taken to avoid impacting the coherence of the collected data. Finally, an ongoing issue is determining and applying algorithms that can “read” for priority in data, as current modes typically scan the entirety of an image with resource-intensive functions instead of establishing a hierarchy of most-to-least importance and operating algorithmically downward.
Transfer Rate and Protocols
Speed is often a counterpart to the size of the data being transferred; smaller, discrete packets can be more readily sent and received. However, it is unavoidable that bandwidth issues present themselves in IoT formats that rely heavily on visual information. Fortunately, investments in network infrastructure (namely, 5G), are helping to ease some of the concerns about network traffic while opening the possibility for new or underdeveloped IoT uses. Priority is an important element in systems with many devices operating simultaneously to determine a flowchart of execution. For instance, a CAN bus sets the protocol and conflict resolutions for communication in automobiles among separate microcontrollers and sensors to elevate signals needing immediate attention. Operating with some level of weighing institutes time efficiency, a necessary component of systems where decision-making needs to be performed continuously in millisecond frames.
A model can only be as valuable as its empirical results. There are two elements to consider: the quality of the data and that of the evaluation. For both, better performance usually comes at the expense of greater resource usage, which may be unavailable or infeasible. Data-rich sensors like high-definition cameras and recorders can provide greater accuracy, but imperil scalability. Performance may suffer without dropping frames, which consequently affects the accuracy of the accumulated data. Similarly, more intensive models perform better, but also require better hardware to avoid impeding the speed of the neural network analysis.
Evolving Needs for IoT Networks
Beyond performance, IoT systems are responsible for data that can be highly specific to the persons or entities that operate alongside them. Historically, these have been of reduced importance to IoT system design. However, the increased cost of compromisation in a digital age where devices are tightly integrated with users makes the case for greater enshrinement.
Designers must be aware of many of the most common ways that system information is improperly accessed by outside agents. Hardware may include bit-enabled security features like writable pages that are rendered non-executable. This requires additional, nonstandard hardware, which may face resistance in devices intended for mass production. A deterrent like randomized address space layouts minimizes susceptibility to brute-force attacks but requires additional memory storage; implementing this style of defensive measure may run per board cost in excess of what design teams are allotted. Traditional methods of securing data may seem to fall apart in the IoT ecosystem, but there exist software and even hardware solutions that work to reduce vulnerabilities. Distributing privileges throughout a network can effectively prevent overreliance on single-point access. Ironically, this stands in stark opposition to a central tenet of IoT systems, but designers can proceed in such a way that a preferred class of devices have access to data deemed sensitive, while other interacting devices do not. Second, microcontrollers can often restrict access to physical memory read/write locations with permissions. Lastly, cloud-based IoT systems can utilize authentication factors as a digital handshake to ensure confidentiality, possibly aided with cryptographic passcodes generated at runtime.
Although cloud computing is often synonymous with IoT due to the ability to offload processes computationally to more capable hardware, some systems are bucking the trend by switching exclusively to edge computing, where device locations are either set or constrained to a small area. For IoT systems that do or will continue to interact server-side for computations, there are a handful of emergent applications. Like with authenticating factors, cryptographic encryption of data can significantly confound efforts to improperly access data. Masking can also be used to remove any potentially sensitive information before the transmission while still retaining useful data to be sent off for processing. Privacy concerns have generally lagged behind those of security, as they tend to affect individual users instead of industry operations, but growing demand will likely continue to shape product marketability.
Cadence Offers Design Solutions for IoT-Enabled Hardware
IoT networking components invoke many different, sometimes countervailing, issues like network accessibility, speed, safety, and more. As with all electronic design, there’s no such thing as a free lunch; design teams will need to collaborate closely to develop a network model that is best suited for the application of the IoT system. Cadence offers support on the hardware end with an all-encompassing suite of PCB Design and Analysis Software tools that can quickly and accurately diagnose a host of design or manufacturing issues even before prototyping. Furthermore, look to OrCAD PCB Designer for a powerful, yet easy-to-use ECAD environment that prioritizes speed in design without sacrificing performance.
Leading electronics providers rely on Cadence products to optimize power, space, and energy needs for a wide variety of market applications. To learn more about our innovative solutions, talk to our team of experts or subscribe to our YouTube channel.