|
Automatic Provisioning Of Broadband Services
A white paper explaining how to “really” automate broadband service
provisioning
By: Bruce Bahlmann - Contributing Author (your
feedback
is important to us!)
Created: December 7, 2001
Note: For help designing your provisioning system or developing tools to help test, automate, and deploy your system contact Birds-Eye.Net.
Abstract
One of the major reasons for the difficulty in automating
provisioning systems today is their belief that they can simply replace employee driven
efforts with technology and complete manual tasks in a fraction of the time. In fact,
automatic provisioning requires technology and innovation to do employee’s tasks
several times better, taking much more into consideration, and increasingly performing
each task more intelligently. This paper will introduce the use of a knowledge base and
the impact it can have on automating provisioning.
Introduction of the Problem
An ever-present feat in the business of delivering broadband services
is that Broadband Service Providers (BSPs) must continue to grow their subscriber base. To
insure profitability during this process, they must begin to streamline subscriber
activation and its technical support obligations by augmenting personnel with technology
wherever possible. Figure 1.0 represents a complex sequence of events that each potential
subscriber must experience. Today, BSPs must rely on its internal employees to process
each phase of the subscriber care process. If
unaltered, this single fact will continue to limit BSP’s ability to scale the
business.
Figure
1.0 Subscriber Care Process
Replacing employees with technology is no simple task. That is
because employees have the ability to think, use the benefit of experience, and can always
call upon a wealth of support options that are difficult to duplicate with technology.
There is also the human factor that any technology centric approach lacks. Employee driven
efforts can be much less complete and accurate than technically driven efforts and will
make up for any of these shortcomings by capitalizing on intangibles such as
friendliness/helpfulness that only personal attention can achieve. This kind of personal
service becomes a differentiator among BSPs and often is a significant driver of
subscriber retention and loyalty.
However, without innovation and automation the workload of service
and support personnel can negatively impact their ability to spend quality time with
subscribers. Employee driven efforts are also extremely inefficient, difficult to
standardize, and above all unpredictable. In fact, employee driven efforts can become
outdated over time as innovation and automation chip away at the manual scope of work.
When this occurs, it becomes difficult to motivate employees to complete repetitive work
that requires increasingly less intelligence and thought. Repetitive work also contributes
to employee turn over which is extremely expensive for the BSP as during this time it has
limited resources until it can hire a new person and then train them.
Translating Human Intelligence into Technology
Automating provisioning must seek to not only complete the basic
tasks required to activate a subscriber but also attempt (where possible) to integrate
some humanized aspects (and/or intelligence) into the activation procedure. While it is
difficult (if not impossible for technology to replace a friendly smile, handshake, and
dazzling personality), technology can over come many of the benefits of having a person go
to the subscriber site.
Figure 2.0 Hierarchy of
Provisioning System Knowledge Base
The key to enabling technology to overcome having an employee on site
is readily available information about the subscriber, their devices, and their relative
environment. For a provisioning system to do
this, it must be able to know more about what is happening at the subscriber site (as well
as within the system of components that interact with the subscriber site) and create a
hierarchy (see Figure 2.0) of information during the provisioning sequence. This hierarchy
of information provides the provisioning server with an advanced knowledge base of the
network, subscriber, operation, and nearly all equipment in the system. Through the use of
this knowledge base the provisioning system can act intelligently on any number of tasks
from the simplest to the most complex with no human interaction necessary.
Like the human brain that can leverage experience and multiple senses
to make decisions on the fly, technology can replicate some of this capability to a
limited but useful extent. Successfully applying the use of a knowledge base empowers a
provisioning system to do uniquely advanced tasks. The scope of these advanced tasks has
the capacity to go far beyond what any one person could possibly perform alone, as
employees unlikely know (or could easily relate) multiple aspects of the entire system in
a timely manner to perform each task. This is where BSPs can actually leverage real-time
data contained within their system to automate subscriber acquisition, troubleshooting,
planning, forecasting, and many important network operations activities.
As service delivery systems become increasingly multi-purpose
they’ll need to expand the number of services combined over the same transport media
and/or technology. With each new service comes a host of new requirements and
specifications that must be met to sustain it. The transports or delivery systems for
these services become increasingly sensitive to change, as it is difficult to keep track
of what all equipment, services, and policies are being used over the common media.
Additionally, as each service grows in maturity it will be all that much more difficult to
integrate them with other services. Automatic provisioning should therefore evolve as a
multi-purpose service provisioning system that addresses operational and business needs of
several services.
Creating and Using the Knowledge Base
Automatic provisioning is permitted by a sequence of events that
together empower an application(s) running in a remote location in the network to
activate, troubleshoot, or service a subscriber. Depending on the amount of information
collected by the application(s) this process could be entirely manual, completely
automatic, or somewhere between these extremes.
While this paper will emphasize a more automatic approach
towards subscriber activation and maintenance, it should be noted that there are several
other ways to do this with varying amounts of employee interaction required. The best
approach is what is right for the BSP, however experience has shown that subscribers
actually prefer having a choice of several different options for installation/activation
as well as service calls. Since “automatic” generally means hands-free, these
other approaches involving increasing amounts of employee interaction will not be
addressed at this time but suffice to say that they do exist and have a useful purpose
within the spectrum of services offered by the BSP. The BSP should keep this in mind and
plan to offer a full range of options to maximize its use of available resources and
technology to best address the needs of their subscribers. The presence of a knowledge
base can positively impact all of these methods.
Performing strictly technical solutions to problems of subscriber
activation, troubleshooting, and service commands a number of technologies namely
self-activation (a.k.a. autoprovisioning, self-provisioning), self-care, and self-service
respectively. These technologies are mainly found deep in the network and require
sophisticated interfaces to various service specific applications, billing, and
operational data sources.
There are also a host of other applications that, depending on the
type of activity, may be invoked. Back in Figure 1.0 a number of things must occur prior
to subscriber activation (Awareness, Selection, Qualification, Sign-up, and Leads
Tracking). Each of these presents an opportunity to collect information about the
subscriber, where they live, what services they want, how they are going to facilitate the
service (using what devices, OS, etc.). This information can be collected from the
subscriber location. While some parts of this data entry could take place outside of the
subscriber’s home (e.g. at a retail store or on the web), at some point it is
necessary for the subscriber to provision/register their equipment (devices in their place
of residence) with the BSP who will be providing them with service. When this occurs the
information contained in Figure 3.0 starts to circulate within the BSPs network.
Figure 3.0 Sample of
Information Available about the Subscriber Site
Figure 3.0 provides a small sample of the kind of information that is
available about the subscriber site. The subscriber site information is perhaps the most
critical as it creates the beginning of a chain of information that links the subscriber
with their equipment and environment. The rich set of information coming from the
subscriber site permits a wealth of options for technology to intelligently empower the
subscriber or for that matter any BSP employee at that location or remote to that
location.
There are four key areas described in Figure 3.0 that dictate the
type of information available at the subscriber site. They are the Dwelling Demark, Modem,
Customer Premise Equipment (CPE), and the Subscriber. The Dwelling Demark is simply the
point at which the service enters the subscriber’s place of residence. This area is
important as it describes a point at which the service becomes unserviceable without the
subscriber’s help – as one would need access to the residence to repair anything
within. It also represents a category in the knowledge base where by everything beyond
that point is associated with the same location/address. The other key areas are
relatively self explanatory other than the CPE which can represent any one of a number of
different devices ranging from a computer, set top box (STB), residential gateway, Media
Terminal Adapter (MTA), or other home appliance.
What happens if this information is not collected or perhaps not
used? Provisioning systems that do not leverage information previously gathered or stored
become one dimensional and rather generic in their treatment or reaction to programmed
tasks. For example, such a provisioning system would simply provide a single template that
all subscribers would follow or traverse no matter what equipment they had or environment
they reside. Any services offered over such a system must be compatible with every
subscriber’s equipment thus it usually matches the lowest common denominator (lowest
tech product in the field). This creates problems as new products are released and the
desire to provide multiple services is then restricted by the lack of intelligence in the
provisioning system to handle advanced services and equipment. It also discourages
subscribers from seeking out advanced devices for their place of residence since the BSP
provisioning system does not recognize such capability nor tailor services to enable the
subscriber to take advantage of it.
Provisioning systems that leverage a knowledge base look much
different. They can make decisions on the fly based on information coming from the
subscriber, their equipment, or the environment they reside. Having this information
readily available enables the provisioning system to customize the experience to the user,
as well as tailor the service offerings to match the actual capabilities of the subscriber
– rather than conventional provisioning systems the only offer a selection of one
size fits all services. In such a system, the BSP is free to offer a multitude of services
each tailored to specific niches of subscribers with the equipment to support those
services. For example a subscriber who just purchased a residential gateway with a built
in MTA could then log on to his account and provision this device in lieu of his/her
existing computer or STB. During this process (if the BSP elected to offer this option) a
new listing of services would be unveiled to the subscriber based on the change that has
occurred with his/her environment/equipment. Subscribers using these types of provisioning
systems are encouraged to seek out advanced equipment because they can now take advantage
of its capability using the intelligent services offered by the BSP. Other services could
also be offered based on address, fiber node, network, city, headend, region, preferred
language, and demographics that would enable BSPs to target services for certain markets,
subscribers, or other criteria.
Provisioning systems that offer this type of “intelligent
offering” leverage something called a selection engine that can evaluate incoming and
stored data using a set of rules that are triggered by the values of the data. Although
both the rules and a majority of the interaction are generally static, the application of
rules in reaction to changing data allows this static content to become highly customized
to the subscriber, their equipment, or environment.
While the information contained at the subscriber location is vital
to the success of automatic provisioning, a number of other areas provide significantly
valuable information that allows even further innovation and customization of services for
the subscriber. Information obtained from these areas must be correlated back to the
subscriber site information, however, before it can be useful.
Information obtained that is beyond/above that of the subscriber
becomes increasingly less subscriber specific. At this the point the subscriber is no
longer directly related to the information. For example, information about the transport
(see Figure 4.0) enables the provisioning system to associate the connectivity of the
subscriber’s residence back to a specific transport media. Without this relationship,
it would be difficult to use information from the provisioning system to inform BSP
service personnel that a particular transport is falling out of specification.
It is important to understand that relationships between data can
work both ways. For example in the case of Figure 4.0 one can determine what transport is
associated with each subscriber – one can also determine which subscriber(s) are
associated with which transport. Both relationships are valid and important as depending
on the perspective of the person requesting information, data relationships should
facilitate one to navigate up or down to find the information they are looking for.
Figure 4.0 Sample of
Information Available about the Transport
Figure 4.0 explains what lies between two demarks in the Dwelling and
the Headend/Hub. Between these two facilities resides the last mile distribution network
for the BSP. A small sample of information is listed however there are actually hundreds
of pieces of information available here.
The information obtained from this area can help the provisioning
system understand the operational environment of the transport that it must use to
activate and maintain end user devices. Knowing the health of this transport can help the
provisioning system determine what its current capacity is – potentially staggering
users on different transports to offset load or until some problem is corrected on the
primary transport. Since broadband transports can carry multiple independent data
circuits, the provisioning system can command which end user devices get assigned which
transport -- if secondary transports are allocated to provisioning. Alternatively, much of
this could be handled by hardware (transparently between the CMTS and modem).
Just up from the transport lies the Headend/Hub Demark. This area
creates a boundary between two different transport technologies. Broadband transports are
typically Hybrid-Fiber Coax (HFC) that delivers one way and two way content, analog
signals, digital signals, etc. However beyond this demark, signals are mainly digital and
traverse an Internet Protocol (IP) network.
It is important to keep in mind that broadband’s last mile is
not inherently fault tolerant. A break in the line or failure of some critical component
can adversely affect its ability to deliver service to those subscribers down stream from
the failure. While the BSP has increased the number of hardened components (by adding fail
over capabilities), a number of them remain single points of failure. They remain single
points of failure as it is cost prohibitive to the BSP to double up each and every active
component in the system for such a small population of its subscribers. To balance the
cost prohibitive nature of making everything redundant, BSPs reduce their exposure to
failure of these components by treating each node separate as it enters the transmission
facility (Headend or Hub). By treating each HFC node separate failure in any one component
of a particular node will not impact other nodes. However, as one traverses back up into
the BSP network each component increasingly serves more subscribers. A majority (if not
all) of these components are redundant as a result.
The lack of redundancy in portions of the BSPs last mile delivery
system create difficulties for network management, troubleshooting, and installation.
Unlike traditional telephony services that use twisted pair to connect up each and every
subscriber, BSPs service heavily relies on the broadcast model to deliver services to all
subscribers on a particular node. While the broadcast delivery methods shares many things
in common with its telephony rival in that each twisted pair has a number of single points
of failure, the BSP must cope with the fact that its single points of failure can possibly
impact a larger number of subscribers than twisted pair. There is also the issue relating
activation to a particular address. Traditional telephony (twisted pair) requires signals
at each address to be activated individually. This not only prevents theft of service but
it also ensures that anything delivered down a particular twisted pair are destine for
only one physical address. BSPs activate an entire node and then wire each home
independently. This method of activation has several advantages over the twisted pair
model. The broadcast model is cheaper to run, uses equipment shared by a number of
subscribers, and supports self-activation for the BSP because their service is
“always-on”. Traditional telephony is not always on and requires a call into the
local exchange to activate service. BSP must manage the risk of theft of service with
their always on model by encrypting content and registering/provisioning hardware.
The challenge for BSPs when provisioning is to know where these
devices are being registered (with which customer, on which node, etc.). Since all BSP
services are being broadcasted to everyone, keeping track of who is getting what and where
they live is critical to their success. However, this is no trivial task. Nothing
inherently relates independent operational data (that which activates devices and services
on the network) back to their physical location. This is why BSPs are dependent on
trustworthy installers to only activate services as directed by their work order. Any
changes to the work performed can result in discrepancies between what the subscriber
receives and what they are actually billed. This also requires BSPs to check in all work
performed just in case anything has changed – so it can be amended in the
customer’s billing record – again, to make every attempt to keep the field in
line with what is in the billing system. More automated systems have evolved since the
early days of broadband services, but many BSPs have been slow to adopt all these
technologies as they still service a generous portion of their subscribers with
labor-intensive analog video services.
While next generation services benefit from being provisioned
centrally, they still lack the backend data links to make this a much more reliable data
exercise. As a result, when you provision something, it can theoretically be active
anywhere across the BSP’s distribution system – an advantage and disadvantage of
using the broadcast model. It’s an advantage because you don’t have to specify
where the device will be installed and it also helps allow technology to transparently
keep the device associated with the subscriber no matter what hardware changes between
these end points. The disadvantage is that the BSP can never be quite sure where the
provisioned and active devices are located in their networks. BSPs must currently rely on
database relationships when correlating subscribers that have signed on with what is
actually manually wired at their location. Without this information, there is little the
BSP would know about what services are being delivered to their subscribers. A more recent
trend for BSPs has been to control what services are being delivered to subscribers by
maintaining the subscribers’ subscription with their account number. Since the
account number also has billing and address information related to the dwelling being
served, this creates similar model to that of telephony’s twisted pair while still
supporting the “always on” model for broadcasted services. Here the
subscriber’s account number becomes the key that links operational data back to
subscriber and dwelling data.
Note that this is only a database exercise as nothing within the
BSPs systems directly linked to the service delivered to a particular address with their
billing information. While equipment such as addressable taps (which can turn on/off
services to specific addresses) do exist, they represent less than 1% of all the BSP
subscriber connections.
Perhaps just as important as creating all these links is the ability
to ensure that little duplication of data occurs. Each piece of operational data can
reside on one or more databases and problems in duplication can occur when trying to keep
everything related. If you try to relate things that can change frequently you run the
risk of maintaining these important relationships. BSPs who do this successfully select
relatively static information such as account number on the billing system and relate this
back to something equally static on the operational data side as a service id (or
equivalent).
The goal here is to automate the process of activation,
troubleshooting, and network management for the BSPs – to make this system
increasingly automatic! Automatic provisioning requires the replacement of manual
intervention with equivalent knowledge bases and innovation. Unfortunately, innovation has
to be at least 300% better than manual efforts to be successful. It is not enough to just
replace a task previously performed by a human with some program that can do the job in
half the time. Automation must do the job consistently better than the human and be
“always accurate”. That is because humans, while inefficient, can confess to
making a mistake and call for help – resolving any problems immediately. Automation
must check and re-check all work as there is no big brother or supervisor watching
it’s every move. If automation makes a mistake it could remain that way indefinitely,
result in significant loss of revenue, and possibly undermine BSP efforts to build
subscriber confidence in their service.
Doing the best job possible from a technology perspective requires
information. Information to make the best possible decisions (programmatically) and
perform the most detailed analysis (or checks) as possible. It would be nice to say that
having all available information would solve this problem and make automatic provisioning
fail-proof. Unfortunately this is unachievable as there are barriers to obtaining
information. Systems seeking the coveted “automatic” qualification face
challenges such as storage, bandwidth, and horsepower when attempting to gather, store,
and use what information is available. Instead, choices/concessions are made to gather and
store the most relevant information possible to achieve satisfactory automation.
Figure 5.0 Sample
Information about the Network
Figure 5.0 goes one level beyond the headend/hub demark to provide
the provisioning system with more information about how the HFC nodes and subscriber
networks are combined. Short of having information about which physical address each
subscriber is connected to, the next best thing is to know which HFC node each subscriber
is connected to. This has long frustrated BSP network operations and troubleshooting
efforts. However to do this requires information about how HFC nodes are combined within
the headend/hub and then connected to the Cable Modem Termination System (CMTS).
Each Headend/Hub maintain connectivity with a large number of HFC
nodes. The number of HFC nodes is dependent on how many subscribers each node serves
(homes passed) and the relative population of the city(s) being served by each facility.
CMTS provide connections for these nodes to connect to it however it is often cost
prohibitive for the BSPs to connect HFC nodes to CMTS in a one-to-one fashion. While a
one-to-one relationship between CMTS and HFC node is simplest and would not require any
further relationship, BSP often combine HFC nodes before connecting them to the CMTS to
save money. When BSPs do this it becomes more difficult to troubleshoot devices downstream
from the CMTS in terms of which HFC node they reside – billing system databases are
also not of much help here as they cannot be fully trusted when it comes to which HFC node
a subscriber belongs. As a result, the only way to know this is either know the transport
(which not many people do), or have some way to know which HFC nodes are combined. While
the later still doesn’t tell which exact HFC node one is on, it does narrow it down
significantly. Mapping each HFC node upstream channel to a single (non-combined) CMTS
upstream interface is critical to enabling automation in determining a subscribers’
HFC node*.
* Note that mapping nodes as described here avoids duplicating
this information in an external database – such as a billing system (how it is often
done today). Billing systems have no direct link to this information and no facilities to
update/refresh it – to do this requires manual intervention. Node information, while
relatively static, does change. When it does, this creates inaccuracies in any database
where it was stored. Instead, node information should be aligned with how the network
hardware is wired and configured and support updates as change occur to ensure 100%
accuracy in this information. Only then can one establish reliable node information that
links subscribers with their associated HFC node.
Figure 5.0 also shows how HFC nodes relate to the CMTS and eventually
to the IP subnets that they are assigned. Tying all this information together from the
subscriber to the IP subnet they belong creates a powerful knowledge base on which to
build automatic provisioning. In this way, automatic provisioning cannot only ensure that
each provisioned device properly joins the network but it can also have visibility to
other devices on the same network. This allows the provisioning system to compare the
newly provisioned device with other devices, ensure its within operational specifications
(of all devices), chart its operational history, and call attention to drastic changes in
the system that have caused chain reactions to devices within its visibility.
Beyond the hardware that facilitates broadband services, lie the
applications and databases that sustain these services and permit important things like
billing, customer care, and expanded service offerings. Figure 6.0 introduces these
elements and what information may be contained within them. While a number of these
applications exist today, none of them are bound together to create a system of
interlocking data that forms a knowledge bases of data linked together for the purpose of
enabling automatic provisioning. Without a knowledge base automatic provisioning is not
possible.
Figure 6.0 Sample
Information Beyond the Network
Figure 6.0 shows the high level applications that enable broadband
services. These applications are tied together through an internal network(s) and
individually perform their designated function (one exception is the Enterprise Product
and Subscriber Management application which does not yet exist) to deliver broadband
services. Today these all operate independently with trivial (if any) links between their
data. The key piece missing that would bring all these systems together is the Enterprise
Product and Subscriber Management application. Broadband operators have a difficult time
standardizing and managing various service specifics such as graphics, logos, advertising,
copyrights, service contracts, definitions, and configurations. These functions must be
managed at a central or enterprise level that then feed to regional (independent)
operating units who actually provide service to subscribers. While these independent
regions also have their own advertising, content, logos, configurations, services, etc.,
they do not compete with enterprise settings but rather compliment it. Merging the needs
of these two entities can be a full time job but when the data is combined it rounds out
the knowledge base.
Automatic provisioning therefore does not describe a service that
once it activates devices remains idle until the next request to provision. This type
activity more closely resembles human efforts that perform a sort of fire-n-forget
approach to installing new subscribers. Automatic provisioning performs more like a
fire-n-remember approach to provisioning. With each job and piece of information collected
an automatic provisioning system must improve/excel. Where employees reach a plateau in
their ability to perform installs or service calls, automatic provisioning must continue
to evolve, innovate.
Perhaps one of the most significant differences between automatic
provisioning and employee provisioning is as employees plateau in their abilities, they
move on, perhaps even leave the company for greener pastures or a new challenge. As a
result, BSPs are hard pressed to continue to improve upon their services
year to year. Instead the best they can hope to achieve is some precarious
state of efficiency that can greatly fluctuate from year to year. However,
automatic provisioning becomes an asset that doesn?t leave the company or go
away. The data that it collects and maintains creates a wealth of
information for demographics, customer care, network planning, service
package design, marketing, and leads tracking that builds year after year.
This increasing, evolving collective of information advances the
organization as a whole and is merely in its infancy in terms of its
industry acceptance and recognized value.
Can Birds-Eye.Net help you or your Company?
Receive your Birds-Eye.Net articles and white
papers hot off
the presses by adding our RSS feed to your reader.
|