In the very near future, sensors will be built into almost everything—there’s even a paint that produces data. Cars, appliances, cameras, doorbells, manufacturing equipment, to name just a few devices, are generating, figuratively, tons of data. Much of this data will flow through the cloud—traveling through the telecommunications (telecom) infrastructure—and creating an ongoing challenge to the telecom companies to provide enough bandwidth. The routing, switching and other tasks that telecom companies will deliver will be based on high performance hardware and other components, and also a change in the way data is handled.
Today’s “connected” cars provide a wide range of functions. In-vehicle entertainment systems and in-vehicle Wi-Fi use a connection to a telecom provider, with cell towers providing the internet connections. Each model year, even more capabilities appear in connected cars—and more bandwidth is required from the telecom providers in order to support the features.
In the very near future, the 5G telecom standard will make it possible for devices to be attached to the internet practically anywhere. With increased geographic range, 5G will reduce the need for nearby cell towers in order to connect devices. With the constraint of proximity easing, the amount of data that will go into the internet will be practically unlimited.
Now, consider that there are hundreds of thousands of these doorbells. Intelligent security cameras, many equipped with speakers, microphones and loud alarms, are also in use. Many of these systems are equipped with facial recognition capabilities.
Consider a camera that monitors traffic on a street corner. The camera can potentially generate megabytes of visual data each minute. Without intelligence built into the camera, the amount of data created by thousands of cameras, on thousands of street corners, can potentially flood the Internet with mostly useless data.
Multiply what’s going on with this simple set of Internet of Things (IOT) applications by the billions of similar connected devices that we’ll soon be seeing, and it’s obvious that the amount of data being moved between devices or through the cloud will be staggering. The amount of data that is being captured (even by passive cameras that send only images, one frame at a time, to a corporate or cloud data center for monitoring), has the potential to overwhelm the telecom systems that provide the bandwidth to support the increasing data flows.
Telecom providers are prepared to deliver the bandwidth that is necessary to handle the data between cloud data centers, corporate data centers and the millions of edge computing devices that are in use now and will be put into use in the future. This will require extremely high-performance hardware and infrastructure to support the routing, switching and other tasks that are essential to assuring the speeds that are necessary to match the flow of data throughout the telecom systems. To support the performance requirements, telecoms are continually adding switching, routing and other hardware that can handle the volumes of data that are passing into—and through—them.
To keep up with performance requirements that are regularly increasing, telecoms use the fastest and most capable hardware. Hardware manufacturers usually provide the equipment that enables high performance, although some organizations are closely involved in design and manufacture of the devices. The telecommunications devices are often based on Xeon CPUs and Intel field programmable gate arrays (FPGAs). FPGAs are semiconductor chips that are designed to perform specialized computations at extremely high speeds. They can be used to accelerate processes to speeds more rapid than a CPU can perform them, boosting the overall performance of the telecommunications device (or any device into which they’re installed).
An advantage that FPGAs provide over other types of silicon is the ability to be reprogrammed quickly so that they can easily switch from one task to another, enabling the device to switch processing functions. Programmability provides flexibility that isn’t possible using other devices.
Even with the increasingly high-performance processing delivered by telecom providers, other steps are being taken to prevent the world from drowning in the data that is generated by the ongoing wave of devices:
Processing Is Being Done ON The Edge Device That Creates The Data. For example, a smart thermostat continually monitors the temperature at the thermostat or attached sensors, but it would only act to control the heating or cooling system when target temperatures are reached. Adding intelligence to thermostats enables programmability so that they can not only set different temperatures at different times and on different days, but they can also sense when the home is occupied. In an unoccupied home, the thermostat may turn the temperatures down on a cold day, raising the temperature an hour before the homeowner usually returns (or when instructed to do so using a remote application). Smart radio frequency identification (RFID) scanners can automatically scan items on store shelves. Intelligence in these scanners can be used to control robotic devices that stock or move scanned items—robotic devices can then move misplaced items to their proper position on the shelf, and can be used to restock the shelves with the right product in the right location.
Data Is Being Aggregated And Only Transmitted When Conditions Are Met. For example, a camera mounted at an intersection would count the cars passing through the intersection. It would total the number of cars (perhaps number of cars per minute) and transmit the total, rather than sending a message when each car passes under it. The RFID scanner mentioned in the previous paragraph can also aggregate the results of a scan of the products on the shelves, transmitting a summary report of the number and location of each item, and sending a report when stock of particular items falls below a certain threshold. This information will be used for ordering products and precise inventory control.
Edge Data Centers Process The Data.
This type of data center (also referred to as a cloudlet) is located near the devices generating data (possibly on streetlights, signal lights, cell towers and other widely distributed locations that are near to one another and to the edge devices), and performs processing of the data sent to it. For example, autonomous vehicles may be sending data to, or receiving data from, these edge data centers. The edge data centers are mounted in locations near the streets or highways, so that it takes only milliseconds for the vehicle to send and receive data from the edge data center. The high performance of the edge data center is enabled by high-speed CPUs, FPGAs and other low-energy devices. Very little of the interaction between edge devices and the edge data centers is sent through the cloud, to cloud data centers or to other data centers. Instead, edge data centers will most frequently process the data that they receive, passing signals to other edge data centers or to vehicles or other devices to which they connect.
In summary, through the use of high-performance hardware and improved infrastructure, telecom providers will be able to handle the data that will be storming through them in the future. This wouldn’t be possible without intelligent management of the data produced, which reduces the amount of data actually put into the cloud. – Forbes