Emerging Trends in Interconnects: From Edge Computing to 5G Networks

As technology continues to advance, the demands on modern interconnects are only increasing, especially with the emergence of edge computing and 5G networks. What challenges do these new technologies introduce? How will connector technologies need to change to cope with advances in technology? What future technology trends will also require changes in interconnect technologies?

What challenges do emerging technologies introduce in interconnects?

In the past, the world of computation was fairly straightforward, with servers handling heavy traffic distributed across large networks, personal computers being used for daily applications such as word and graphics processing, and mobile devices being used solely for mobile communications. The same also applies to the first IoT devices. These were primarily basic microcontrollers with limited internet connectivity, able to provide routine readings from a few basic sensors or receive commands to perform actions (such as opening doors and windows).

However, over the last two decades, the nature of technology and how it is used has fundamentally changed, with the lines between types of computational devices blurring. For example, mobile devices that were used only for communication are now able to perform numerous everyday tasks including email, web browsing, and even video editing — making them almost as capable as mainstream desktop machines.

Image 1: The dramatic change in technology has put massive pressure on interconnect designers

Another example of a major technology change is the introduction of edge computing, whereby data is processed on-device instead of being sent to a remote server. A major benefit of edge computing over the use of remote servers is that private data can be kept on the device without ever being exposed to the outside world. Applications that can benefit highly from edge computing include face-detection camera systems, wake-word systems, and medical sensors.

When looking at edge computing from the perspective of a network, the reduction in bandwidth usage helps to minimize congestion while allowing for more devices to connect to a single network without suffering from major performance issues. However, the increased local processing needed for edge computing introduces a multitude of challenges for devices, including the need for increased power, reduced battery life, and increased device complexity.

But it’s not just IoT devices that are seeing a move toward edge computing — even 5G installations are actively deploying such technologies. The ability to run essential services locally on a cell network not only reduces the overall internet traffic consumed by the cellular network but can greatly reduce latency.

Fundamentally, what all of these changes in technology translate to is that modern devices have large high-speed data requirements. This requires high-speed data ports as well as board-to-board connections where applications need to deploy a degree of future-proofing; some modern designs are exploring the use of System-on-Modules (SoMs) that can allow hardware to be upgraded after installation.

For example, modern IoT devices are no longer trivial microcontrollers with a few kilobytes of RAM and a small CPU. Instead, they’re full-fledged systems with hundreds of megabytes of RAM, advanced CPU features, and, in some cases, AI coprocessors to help accelerate the speed at which AI tasks are executed while reducing energy consumption.

However, interconnects in emerging fields are not just required to provide high-speed data capabilities, but also to allow for the creation of ever-smaller designs (miniaturization) while supporting higher power levels. This introduces a whole range of challenges for engineers when designing interconnects, often resulting in all kinds of innovative interconnect designs and options.

How will connector technologies change to cope with these emerging trends?

Undoubtedly, the first and most important change that interconnects will have to undergo is the support for larger bandwidths. While interconnects already exist for bandwidths exceeding 100Gbps, they are only ever found in extreme applications, such as backplanes for data centers and supercomputers.

Now, it is highly unlikely that future edge-computing applications will see a need for such high data rates, but having the ability to operate at such speeds not only allows for massive amounts of data to be processed, but also helps to future-proof designs so that eventual upgrades do not need interconnects to be replaced. But how can interconnect systems support such tremendous bandwidths in edge-computing applications and miniaturized cellular network controllers?

The J67 Series is ideal for high-speed data center applications

One option for supporting higher bandwidths in interconnects is to increase the number of poles, especially ones that support differential signals. However, simply adding more poles isn’t a straightforward solution, as this also increases the size of the resulting connector — thus, pole density also needs to increase (i.e., more connectors per mm).

But making interconnects smaller introduces its challenges, including increased issues with EMI as contacts become physically closer to each other, and reduced mechanical strength. Furthermore, reducing the size of interconnects also reduces the voltage and current ratings, which can be problematic in applications such as GPUs where power consumption can be excessively high.

One potential solution to solving this challenge with increased interconnect densities is to split up designs across several self-contained modules that support local high-speed data processing and utilize slower busses for communication between modules. This style of construction is slowly being adopted by engineers, with one famous example being the Apple iPhone, which separates multiple system components into individual boards.

The 470 Series is ideal for high-speed board-to-board applications

Unfortunately, for all the benefits that modular designs present, they do have their setbacks, including significant latency issues between modules, reduced bandwidth between modules, and decreased overall signal integrity. For example, M.2 Key interconnects are becoming popular within IoT designs, but they are generally used to add additional connectivity options to a pre-existing design and rarely used to rapidly transfer massive amounts of data from one module to another.

Another challenge interconnects face is the need for future-proofing in next-generation designs. An excellent example of this issue can be seen in emerging cellular technologies, whereby hardware used in the underlying mobile infrastructure is shifting toward field-programmable gate arrays (FPGAs); they allow for hardware updates to be reflected in software (thereby allowing fundamental changes, possibly supporting future standards).

However, while the hardware itself may be upgradable, the interconnects used cannot be upgraded. Plus, their physical limitations will impose a hard limit on the overall design. Thus, programmable systems designed to be future-proof need to consider every interconnect used, from the board-to-board module card connectors to the ethernet ports used to communicate with other networked devices.

What future trends will necessitate changes in interconnects?

Going forward, one future trend where interconnects will see major changes is the vehicle-to-everything space (V2X). Interestingly, one of the main goals of 5G was to provide vehicles with a low-latency, high-bandwidth internet connection to allow offloading of complex self-driving tasks while simultaneously providing a network for cars to share data with each other (such as location, direction, and speed). However, the trouble facing 5G — including its slow development and rollout — has seen V2X yet to materialize.

Regardless, should V2X become a mainstream technology, then interconnects in such applications will not only need to be able to provide substantial bandwidth capabilities, but also be resistant to vibration, shock, and extreme temperature swings. Furthermore, as such systems will likely deploy some form of self-driving, safety will be the number one concern for engineers; thus, interconnects will not be able to tolerate momentary disconnects. This need for large bandwidths will arise not just from the use of edge computing and complex AI systems, but the vast number of sensors including cameras, microphones, ultrasonic modules, radar, and LiDAR.

On the topic of V2X, interconnects will also become a critical part of telematics due to the increased use of sensors in vehicles, the use of internet-enabled connectivity, and the rise of self-driving applications. As telematics involves the data processing of sensors in real-time while also feeding this data to other systems, the need for reliable connections that can support high bandwidths will be imperative. Furthermore, as self-driving systems and V2X will be required to produce accurate telemetric data on vehicles, there will be a strong requirement for dead-reckoning systems, which in turn require multiple measurement systems working cooperatively. As such a feature is safety-driven, high reliability will be an absolute priority.

The rise of smart cities will also see an increase in demand from high-bandwidth interconnects due to the increased use of sensor platforms to monitor traffic, the environment, and the population in general. As such, these connectors will need to be able to withstand outdoor environments, or, at the very least, survive temperature swings and changes in humidity.

Overall, the many emerging trends in the electronics industry continue to put pressure on interconnects, requiring larger bandwidths, smaller sizes, and improved reliability. However, the introduction of future-proofed designs now further complicates the task of interconnect selection.