The Internet of Things (IoT) and edge computing offer a wealth of opportunities to communications service providers (CSPs)—but only if they can turn the corner on monetizing their big data.
Many still struggle to reap the rewards of running advanced analytics on their volumes of big data because of ill-defined business goals, data silos, operational fiefdoms and a variety of other reasons.
CSPs unable to resolve these and other issues risk missing the opportunities afforded by IoT and edge computing that will help them remain profitable. While the issues and advice in this article are aimed at CSPS, the principles – along with the opportunities and dangers of inaction – apply to leading-edge enterprises in many industries.
Real-Time = New Revenue, Improved CX
Edge computing is a necessity for minimizing latency and
enabling actions and decisions based on the freshest data available. These are
key requirements of many IoT applications. CSPs already have massive, built-in
edge computing infrastructures in the form of many scattered points of presence
(PoPs), central offices, server rooms, cell towers, base stations, utility
poles, rooftops, consumer homes, and countless other places.
These coveted edge locations can be leveraged to host IoT applications for third parties, or to get into IoT businesses of their own, such as smart homes, utility monitoring, and video surveillance for public safety. Companies can also run real-time edge analytics to improve customer experiences (CX); instigate new kinds of subscriber marketing; combat malware and fraud; sell advertising based on real-time consumer data; and improve network management, operations, and uptime.
Analyze before Storing
Cashing in on these opportunities means approaching analytics in new ways. In contrast to traditional business intelligence platforms used for historical trending, for example, analytics conducted at the edge deliver continuous intelligence with near-zero latency. CSPs, then, can harness newly generated data earlier, in a distributed and naturally scalable manner, before data streams become data torrents.
This is a smarter alternative to hauling highly distributed data over a wide-area network and processing it in a centralized data center or cloud location. Data rates are growing at about 50 percent per year, while costs in storage systems and processing hardware are dropping by only about 15 percent per year. It’s clear that something has to give for CSPs to remain profitable.
See also: From Real-Time to the Edge: The Top IoT Trends for the Coming Year
The cost issue alone is helping drive edge processing
adoption, because it allows unneeded or irrelevant data to be filtered out
instead of being sent over the network and stored, which comes at a significant
cost as data volumes continue to mushroom. Most important, though, edge
computing and analytics—and, increasingly, streaming analytics—are all about
acting upon the most current data for best business value, when time is of the
essence.
Time-Sensitive Use Cases Call for SQL Streaming
IoT edge use cases that rely on
continual, real-time data are as diverse as fraud detection, medical
monitoring, self-driving vehicles, gaming, environmental monitoring, and
augmented and virtual reality. These applications require the lowest possible
latency and must combine edge computing with streaming data analytics to enable
a number of capabilities.
• Process queries in milliseconds.
Streaming analytics platforms continually perform queries on streaming data
across any number of external data sources and enable applications to integrate
components of that data into an application flow. The platforms can process
tens of millions of events per second, which means they can immediately and
automatically collect, correlate, analyze, and act on network and customer data
in diverse formats across systems within seconds of an occurrence.
• Correct issues before they impact the
customer. Thanks to the fast, voluminous processing, CSPs can analyze
their operational, customer, and business data anywhere in the network dynamically,
so they can make better decisions, provide targeted new services, and reduce
costs. For example, they can detect and correct
network issues before they have a chance to negatively affect customers. If
artificial intelligence (AI) or machine learning (ML) components are bundled
into the edge capabilities, the corrective actions can be automated.
• Discover
opportunities for personalized subscriber marketing. If a mobile
network operator tracks a subscriber’s movements across cell sites, the
operator might deduce that the subscriber is about to leave the country and
might be planning to disable the data-roaming feature on his mobile phone to
avoid high international roaming costs. The operator might push an offer of a
special-rate international plan to the subscriber with far more attractive pricing.
Such a move would provide the subscriber with a valuable service while creating
incremental revenue for the operator.
The automated actions and decisions
enabled by edge computing and streaming analytics provide situational and
contextual awareness, which are key enablers for prediction and prevention. Let’s
look at a few use cases.
Use Case 1: Predictive
Analytics
· Preventing network downtime. Today,
it’s common to perform centralized data analytics to determine what has caused
network downtime after it occurs. In a streaming analytics world, an edge
solution could help network operators predict an equipment failure or a network
outage before it happens, allowing them to take action to prevent downtime and to
avoid the sub-par subscriber experiences resulting from downtime that can lead
to customer churn and low mean opinion scores (MOS).
· Improved CX. Real-time subscriber monitoring empowers CSPs to optimize each CX based
on both real-time situational analysis and predictive analytics. For
instance, if a mobile CSP can predict a potential cell failure due to severe
weather conditions that will likely impact a large volume of customers, the CSP
can use streaming analytics to automatically reroute calls to an alternative
cell. Subscribers remain blissfully unaware that a switch was made, only
perceiving that their connections remained intact.
· Public safety. Emergency services are benefitting from streaming analytics’ continuous
intelligence. For example, ECaTS, a cloud-based SaaS analytics platform for
public safety analytics, has combined its system with SQLstream Blaze (acquired
by Guavus) to enable 911 operators to take real-time actions that include the
following:
- Assess incident severity in real-time based
on call type and geography-based call volume.
- Address workload issues immediately and
prevent escalations using real-time call status metrics that are updated many
times per second.
- Immediately identify and rectify network
performance issues that could impact 911 call handling.
· Combatting fraud. The primary challenge around preventing fraud
and network intrusions is the speed at which network infiltration occurs.
Because break-ins can occur quickly and have devastating consequences, it’s
critical that CSPs can both detect and stop fraud as it’s happening and also use
predictive analytics to prevent it in the first place.
· Billing and usage. With
the ability to calculate a subscriber’s usage in real time, it’s possible to
dynamically offer subscribers reaching their monthly data or voice limit the
option to top off their accounts for the month with additional usage. CSPs
might reward those customers by reducing tariffs for voice and data as their usage
increases.
Use Case 2: Monetizing the
Mobile Edge
Mobile operators can
capitalize on computing at the edge of the cellular
network, a function called multi-access edge computing (MEC). MEC brings IT services and cloud
computing capabilities to the radio access network (RAN) segment of the
provider’s network, in close proximity to mobile subscribers. Implemented in cellular
base stations or other edge nodes, MEC enables flexible, rapid deployment
of new services for customers. Operators can also open the RAN to a new
ecosystem and value chain that might include third-party application and
content developers.
When mobile operators are able to track the activity and
movements of their subscribers at an individual level in real-time, for
example, they’re in a great position to offer micro-targeted, personalized
offers, such as the international data plan mentioned earlier. By
definition, mobile operators are aware of where their subscribers are for
billing and emergency services purposes, so it follows that other, creative
location-based services are a natural service type to push to their
subscribers.
The European Telecommunications Standards Institute (ETSI), for example, notes the following about augmented reality
applications: “Hosting [them]
on a MEC platform instead of in the cloud is advantageous since supplementary
information pertaining to a point of interest is highly localised and is often
irrelevant beyond the particular point of interest.”
Use Case 3: Smart Meters, Cities, Homes, Cars…
As CSPs deploy smart metering, smart homes, connected car, and
eHealth applications, the need for streaming analytics alongside more
traditional offline analysis tools becomes even more pressing. Increasingly,
IoT applications depend on applying continual analytics to constantly streaming
data to deliver benefits such as:
- Saving an organization piles of money through vastly
improved operational processes, faster troubleshooting, and new efficiencies.
- Lucrative, fast decision-making based on the
data “of the moment,” such as executing a financial trade.
- Improving a person’s health and even saving
lives by monitoring real-time health vitals and triggering an action—such as dispensing
a medication or dispatching an emergency response team—when necessary.
- Improving public safety by optimizing emergency
dispatch services and processing real-time apps like bodycams worn by peace
officers and facial recognition systems built into surveillance cameras.
Implementation Recommendations
There are a number of issues to consider when evaluating analytics for edge deployment. Below are three tips that may help you successfully advance your edge processing and analytics initiatives:
1.Seek extremely high efficiency
Edge computing and analytics require extremely efficient, high-performing systems, because edge servers are lightweight devices by design. They’re built that way because CSPs must run large numbers of them in their network infrastructure. Be aware that C++ based platforms offer superior processing efficiencies over Java-based systems. Also, it’s important to seek a compact binary data advantage. Compact binary systems consume far less memory to store and represent the same information that other systems would use, which translates into data copying and analytics that’s faster by at least an order of magnitude.
2. Consider the level of expertise you have in-house and how quickly a system will allow you to get up and running.
Expertise is becoming a significant bottleneck to edge deployments. Operators must be able to implement new systems quickly, without having to hire a new team for every new project and without having to scour the world for rarified skillsets, which often come at a high market cost. Consider systems built using proven and familiar SQL standards that can process active live streaming queries while retaining their familiarity to millions of loyal users. These considerations translate into greater productivity, which, in turn, translates into competitive advantage in total cost of ownership (TCO) and time to market.
3. Jump in, even if that means starting small.
Analysis paralysis can often stop companies
from moving forward in a timely fashion with technologies that are unfamiliar
to them. With edge analytics, any
starting point is a good one, because you’ll quickly learn about the
capabilities, performance, and benefits of an edge analytics platform. You can
usually start to build and deploy simple solutions in a matter of a few days
and, from there, additional projects become easier.
Not Rocket
Science
Real-time and stream processing don’t have to be difficult. While distributed parallel processing might sound like rocket science — a clearly defined business goal and a scalable platform that’s interoperable with a wide variety of data formats will help you quickly demystify these systems with prototypes. From there, you’ll be able to decisively move forward with more ambitious and value-generating projects that address your biggest business challenges.
