Monday, September 30, 2019

Top IoT Use Cases Vary by Industry, Study Suggests

In manufacturing, the top use cases for IoT are: automation (48 percent), quality and compliance (45 percent), production planning (43 percent), supply chain logistics (43 percent), and plant safety and security (33 percent), a study conducted for Microsoft by Hypothesis Group finds. 

For retail/wholesale companies, IoT is highly relevant for supply chain (64 percent) and inventory optimization (59 percent), while for transportation and government organizations equipment management and safety/surveillance are particularly important (about 40 percent to 55 percent). Within healthcare, IoT helps companies track patients, staff, and inventory (66 percent), as well as assists with remote device monitoring and service (57 percent). 


California's Rules on IoT Security Devices

A new California law sets mandatory internet of things device security rules. So if suppliers must build to meet the law in that state, they will likely sell the same device everywhere else. 

The new law takes effect on January 1, 2020. In contrast to existing California data privacy laws protecting personal information, the new law aims to protect the security of both IoT devices and any information contained on IoT devices.

The law requires a manufacturer that sells or offers to sell a connected device in California to equip the device with a reasonable security feature or features, including:
  • Appropriate to the nature and function of the device
  • Appropriate to the information it may collect, contain, or transmit
  • Designed to protect the device and any information contained therein from unauthorized access, destruction, use, modification, or disclosure."

While the law only vaguely defines the term "security feature," it requires security whenever a connected device can be authenticated outside a local area network. Well, in an age of cloud computing, that is virtually all devices. 

Reasonable security features include:
  • The preprogrammed password is unique to each device manufactured"
  • The device contains a security feature that requires a user to generate a new means of authentication before access is granted to the device for the first time."

So what is a “connected device?” "Any device or other physical object that is capable of connecting to the Internet, directly or indirectly, and that is assigned an Internet Protocol address or Bluetooth address." 

Sort of any device, in other words.

Sunday, September 29, 2019

Is Edge Computing Going to Dovetail With Wi-Fi Offload?

“AI is very good at consuming large amounts of data,” especially at the edge of the network, says Jim Thompson, Qualcomm CTO said. And that is one reason the value of 5G will in many instances be a combination of ultra-low-latency access, plus edge computing, plus applications using artificial intelligence that must process lots of data very quickly, nearly in real time. 

In the 5G era there might be many cases where edge computing is not strictly necessary to support latency requirements, but might add value as a way of reducing wide area network transport costs, preserving privacy or boosting security. 

The edge cloud might reduce network transport costs and load by conducting some processing at the edge, without transiting over the wide area network. 

Qualcomm believes there’s a “partition between the cloud, the edge, and then this very close edge cloud that is milliseconds away from the device,” Thompson said.

AI might therefore be one of those developments--other than the numbers of smartphone accounts, the amount of data consumed by smartphone users--that shapes network demand. 

And that, in turn, might shape use of Wi-Fi and mobile networks. 

Some have speculated that Wi-Fi offload will be less relevant after 5G networks are in place, especially in markets where unlimited usage or big buckets of usage are the norm. 

But researchers at Cisco believe the huge amounts of data 5G will enable (faster networks have always lead to higher data consumption in any unit of time), combined with usage-based data allowances, will still create value for offloading device data consumption to Wi-Fi.

Perhaps the same might be said of how edge computing winds up being used. Applications will offload to edge computing facilities as a way of reducing the cost of processing huge amounts of data in near-real time or real time. 

The amount of traffic offloaded from 4G was 57 percent at the end of 2017, and it will be 59 percent by 2022, Cisco predicts. 

The offload percentage on 5G might be as high as 71 percent by 2022. One might speculate that offload to local computing also will be part of this trend, especially if edge computing happens on the premises, on the campus or within the metro area. 

Figure 17.    Mobile Data Traffic and Offload Traffic, 2022
Note: Offload pertains to traffic from dual-mode devices (excluding laptops) over Wi-Fi or small-cell networks.


Amazon One-Day Delivery is Like Edge Computing

Amazon provides an analogy to key underlying drivers in many parts of the telecom business, where it comes to shaping demand and supply of connectivity services. 

What Amazon and many connectivity suppliers share in common:
  • Sell retail products
  • In competitive markets
  • With lowish margins
  • Using infrastructure and networks
  • Where volume (bandwidth) is an issue
  • Where latency (delivery time) is an issue
  • Recurring revenue (repeat purchase) is key
  • Sells high-margin and low-margin products
  • Revenue sources have changed over time
  • Regulation is, or is becoming a key business constraint

Consider the issue of latency, or time to delivery. Amazon is boosting its use of local warehouses, to support its one-day delivery services. 

Amazon is already capable of offering same-day and next-day delivery to 72 percent of the total U.S. population, including almost all of the households (95 percent or more) in 16 of the wealthiest and most populated states and Washington, D.C., according to RBC Capital Markets.

That is supported by a network of Amazon fulfillment centers across the country.

The coming analogy in the connectivity business is edge computing, which will support latency-sensitive applications that cannot rely on distant hyperscale data centers. As consumers increasingly prefer same-day or one-day delivery, in preference to two-day deliveries, so many internet use cases and applications eventually might require extreme low latency that only can be supported by local computing. 



Friday, September 27, 2019

5G and Edge are Inseparable, CEOs Say

Majority of IoT Connection Revenue to Come from Enterprise

Ultimately, about half of IoT connections supplied by mobile operators will come from enterprise customers, while half comes from consumer use cases, Mobile Experts aregues. But revenue will come disproportionately from enterprise use cases, as historically has been the case for telecom operator revenue sources. 

Analysts at ABI Research also agree that enterprise applications will drive much of the profit upside. 


Applications for industrial markets will drive much higher revenue than simple consumer devices, for example. According to Mobile Experts, Edge Computing platforms and Cloud platforms for non-real-time applications are now the critical part of developing the IoT market.


"Many people assumed that massive growth in IoT was going to happen quickly and easily. Three years ago, we disagreed, and predicted a slower rise of IoT adoption,” said Madden. “We expect the installed base of IoT devices to achieve some healthy growth from three billion in 2019 to 8.5 billion in 2024." 

Hard to Separate Edge Computing from 5G and AI

Edge computing, 5G and artificial intelligence are fundamentally inseparable, many would argue. The reason is that the advantages of ultra-low latency and ultra-high bandwidth apps and use cases hinges ultimately on use of all three underlying technologies, plus use of the new 5G virtualized core networks. 

Though 3GPP Release 15 standards are aimed primarily at supporting consumer smartphone use cases, Release 16 and 17 will extend 5G frameworks to private networks, IIoT (industrial IoT), and Internet of Vehicles (IoV), according to Jim Thompson, Qualcomm CTO. 

Those new enterprise applications will often require AI and edge computing, he said. 

Friday, September 20, 2019

Societal Impact of Industrial IoT




Dr. Theodore (Tod) Sizer, VP of Smart Optical Fabric and Device Research, Nokia Bell Labs, talks about benefits of
large-scale industrial IoT and IoT impact on a larger social level.

Virtual IoT Networks: Friend or Foe?

In an age where virtual networks will be routine, where specific use cases and apps are supported, what is the potential impact on connectivity service providers? The obvious analogy is the impact of over-the-top apps running on any internet connection in the consumer space.

In other words, when an app or service provider can use IP networks--virtual or physical; on a dedicated virtual private network, using public internet access or a network slice--value and potential revenue shifts from one supplier to another. This is not new, but will be a bigger issue in the 5G era, as VPNs optimized for specific apps may proliferate in enterprise and consumer segments of the market. 

There are both retail and wholesale opportunities, as always is the case for VPN-based networks and platforms. 

Emnify’s CIoT Mobile Core Network is a virtual network is serving more than 1,000 enterprise customers with mobile core network functions including HLR / HSS, GGSN / PGW, SMSC, USSD, PCRF,  STP, DRA, including OCS, OSS / BSS support functions such as billing and provisioning.

EMnify's network supports all cellular technologies from 2G, 3G, and 4G as well as low-power wide-area network (LPWAN) technology NB-IoT and CAT-M1.

Similarly, Telna provides global wholesale networks optimized to support IoT. Monogoto, 1NCE suppliers the IoT platform sold by Deutsche Telekom, which also suppliers the wholesale connectivity for 1NCE. 
Soracom also sells virtualized networks, featuring a for pay-as-you-go data service with connectivity prices as low as US$0.02 per megabyte.

Thursday, September 19, 2019

Edge Analytics Follows Shift to Edge Computing

As data storage and computing shifts to the edge, so does analytics. A survey conducted by Harvard Business Review found 25 percent of respondents saying they presently use some form of edge or internet of things device analytics, growing to perhaps 30 percent by about 2022. 

Some 54 percent of the organizations surveyed have plans to increase the amount of data they store in the public cloud over the next year, but the majority still manage much of their data on-premises, the survey found. 



Wednesday, September 18, 2019

Will Multicasting Cannibalize Content Delivery Networks?

Edge caching and content delivery networks have been used to improve end user experience for a variety of unicast (one user interacting with web apps) and multicast (many end users interacting with a video stream at the same time) use cases.

What remains to be seen is how new 5G core network capabilities--ranging from low-latency access to edge computing to virtualized networks--will compete with more-traditional content delivery network services.

Live game streaming (esports), where viewers watch other gamers compete with each other, is said to be one possible application for network slicing. The notion is that creating a network optimized for point-to-multipoint (multicast) content delivery benefits from a network optimized for low latency and high bandwidth. 


Of course, there are other ways to support such networks. Traditional live video networks, including those featuring 4K content, have relied on satellite transport to local hubs (cable TV headends) instead of unicast delivery. 


It is not so clear that multicast gaming feeds require transport that is materially different from live broadcast (multicast) video, though the intended display screen is PC screen. So the constraint might not be the wide area delivery network itself but the business arrangements around episodic use of the network. Is the content delivered as a consuming-facing “channel” that is programmed “all the time,”  or as an episodic event more akin to a podcast, videoconference or other discrete event. 





Multicast ABR, for example, is a new multicast format proposed by CableLabs for multicasting video content instead of the more capacity-consumptive unicast method of delivery.  

Satellite networks might still work for esports packaged as a live channel. Other WANs might work--especially when edge caching is available, for more episodic events. 


Many say 5G will be an enabler for live game streaming, and that is true in some fundamental senses: the 5G network will use a virtualized core network that makes network slicing possible. 


It also is true that the 5G core network is designed to support distributed computing, and is therefore also an enabler of edge computing and caching, which might also be an approach for supporting live gaming streams, as content delivery networks have been used to speed up app performance generally. 


On the other hand, 5G as an access technology might, or might not, be necessary. A custom VPN is one approach. But satellite delivery to edge locations such as headends also is an option. Multicasting using a network slice also works. 


In that scenario, 5G latency performance might contribute to the experience, but it really is the edge computing or the network slicing that contributes most to the low-latency performance. 


Also, creating a low-latency, high-bandwidth network for the actual playing of esports games is a different matter than streaming of such matches. The former requires a high-performance unicast network, the latter might, or might not, rely on such a network, as the latter is a multicast operation. 


Where most present internet operations are unicast, one-to-one sessions, streaming of video or live esports content is multicast, many-to-many or one-to-many operation. 



WAN latency performance is key, though it typically is not so much the WAN performance as the capabilities of the IP networks using the optical WANs, that dictates the limits of experience. Also, large venues are needed for such competitions, so premises networking and high-bandwidth access facilities are a must. 


The ability to handle episodic surges in the actual gaming might be an issue for the access connection or the WAN transport. That is an issue either network slicing or raw bandwidth provisioning might  support. 


The streaming of selected portions of the gaming competitions is a separate matter. 

Streaming live esports, in other words, is one networking problem. Supporting an esports tournament is another issue. And streaming such competitions arguably is a third issue. Network slicing is one potential way of handling the streaming. But there are likely going to be other ways, as well.

Tuesday, September 17, 2019

Edge Processing Driven by Real-Time Analytics

While data storage is likely to remain “in the cloud,” processing is likely to move increasingly to the edge, driven by real-time apps and use cases, especially those related to internet of things apps.  As much as 45 percent of data created by IoT will be stored, processed, analyzed by edge computing, according to IDC researchers. 


A reason for the amount of edge computing is the shift to “real time” analytics, which in many cases means remote processing delay is undesirable. 


While data storage is likely to remain “in the cloud,” processing is likely to move increasingly to the edge, driven by real-time apps and use cases, especially those related to internet of things apps. In most cases, those IoT apps will be deployed by enterprises. 



Friday, September 13, 2019

AIOps for Anomaly Detection Might be Used by 42% of Enterprises by 2021

Within the IT operations and monitoring space, AIOps is most suitable for appli­cation performance monitoring, informa­tion technology infrastructure management), network performance monitoring and diagnostics and information technology event correlation and analysis, say analysts at Boston Consulting Group.

The common denominator for AIOps is that the systems help automate routine manual operations activities.

 Three use cases hold the greatest short-term potential, according to BCG:
  • Anomaly detection
  • Noise reduction
  • Triaging and alert correlation

Anomalies and unusual behavior—such as a sudden spike in application use—is a clear use case, reducing the amount of manual observation and hard-coded rules that require teams to define anomalies up front. BCG says only about 11 percent of CIOs currently use anomaly detection AIOps tools, but that figure should grow to 42 percent by 2021.

Machine learning algorithms built into AIOps platforms could prioritize alerts on the basis of their business impact and filter out false positives, freeing IT operations teams to spend their time addressing critical alerts instead of managing static filters, writing rules, and adjusting thresholds to reduce alert noise, BCG argues.

About nine percent of CIOs now use noise reduction tools in AIOps, but our survey data shows that the percentage could rise to 42 percent by 2021.

AIOps also could automatically associate alerts that cut across various IT services into a single incident to speed up triage. That could help teams determine whether different alerts are related, and then cluster the results into a single, unified incident. 

For example, a monitoring tool might create multiple memory and page-fault alerts from hosts of the same SQL cluster. The ML algorithm in the AIOps tool, when properly trained through supervised learning, could correlate alerts into a single incident, allowing the IT operations team to distinguish between alerts belonging to that incident and similar but unrelated alerts. 

About 10 percent of the CIOs surveyed by BCG say they already use some sort of AIOps-enabled triaging solution today, and 40 percent say that they’re open to using this type of solution within the next three years.

The market for core AIOps is projected to grow from $9.4 billion in 2017 to $13.8 billion in 2021, a compound annual growth rate of 10 percent. AIOps orchestrators—platforms built to orchestrate insight and actions on the basis of log data from various monitoring solutions—are expected to grow by 26 percent over the same period, BCG predicts. 


BCG believes AIOps will help transform IT operations in three critical ways, providing end-to-end visibility, providing evidence-backed insights and recommendations and executing recommendations automatically.

Eventually, IT teams will have to decide how much they allow the machine learning algorithms to make independent decisions about configuration of services and environments. Few will be comfortable, early on. Later on, with experience, attitudes could change. 



Device Connections Shifting to the Edge

Many observers believe that the overwhelming number of internet of things devices in use by 2024 or 2025 will use some short-range form of connection, such as Bluetooth or Wi-Fi. That does not directly correlate to the location of data processing, which might be on a connected device (smartphone) local premises device, regional data center or far-end remote data center.

But the shift to local connections for IoT does at least suggest a growing role for edge computing of various types.

By 2024, IoT device markets will represent 31 percent of total Bluetooth and 27 percent of Wi-Fi device shipments, up from 13 percent and 10 percent respectively in 2018, say researchers at ABI Research. 

Smartphones will continue to be important markets of strength for both Bluetooth and Wi-Fi , but Bluetooth shipments for the IoT market will overtake the smartphone market for the first time in 2024.

Ericsson forecasts mirror such expectations, showing the greatest unit growth in the connected device market coming from short-range IoT. 

Smartphones as a proportion of Wi-Fi device shipments will fall below 40 percent by 2024. 
“ IoT is beginning to take an increasingly significant share of the market,” says Andrew Zignani, ABI Research principal analyst.

Key IoT opportunities for Bluetooth will be found within asset management and location services in devices such as beacons and personal trackers. These are anticipated to grow from around two percent of the Bluetooth market in 2018 to over 8.5 percent by 2024. 

Analysts at IoT Analytics agree that short-range connections will dominate IoT connections by 2025. 

Bluetooth-enabled wearable devices are also expected to break the 400 million device barrier by 2024, with increased traction in smartwatches, activity trackers, smart clothing, and hearables, the firm predicts. 

Wi-Fi-enabled wearables are also expected to reach over 250 million units by 2024. 

The smart home will be one of the quickest growing markets for both Wi-Fi and Bluetooth technologies. Wi-Fi-enabled smart home devices are expected to grow from five percent in 2018 to nearly 16 percent by 2024 and Bluetooth will rise from four percent to 13 percent by 2024, with traction in voice-control front ends, smart appliances, smart lighting, sensor devices and video cameras, ABI Research estimates.