The Quest for Wireless QOS within a Network Neutrality Framework
1. The Quest for Wireless QoS within a Network Neutrality Framework (Disponible
en Español)
Posted by jormarguo Jul 10, 2012
So we are approaching that time of the year in which Internet Net Neutrality debate regains attention
and media coverage, in this post entry, I share my opinion of the subject with an emphasis on QoS for
wireless ISP. Three sources serve as foundation to develop my argumentation; BEREC (Body of European
Regulators for Electronic Communications) documentation on the matter. FCC Open Internet rule and
the correspondent analysis constructed by Prof. Barbara Van Schewick.
Stepping Ground
In 2009 FCC started its proceedings for the Open Internet, in 2010 rules where adopted which became
effective in 2011. On May 25th this year, the FCC announced the composition of Open Internet Advisory
Committee (OIAC) formed by various actors of the industry whose purpose is to evaluate the effects of
Open Internet rules, to assess the evolution of the industry since the rules were promulgated and lastly
to consider the importance of wireless 4G technology adoption in the American market. The BEREC
started in 2010 its consultation to explore the regulatory aspects of Net Neutrality, by 2011 defined the
key issues to be developed: Transparency, QoS requirements, Discrimination and IP interconnection and
by 2012 several draft documents have been produced under this effort; A Framework of QoS in Scope of
Net Neutrality, Guidelines of QoS in Scope of Net Neutrality and IP Interconnection Assessment in Scope
of Net Neutrality, all available for public consultation at the website. Finally the ITU has published the
related agenda framed in the WCIT (World Conference on International Telecommunications) to be held
in Dubai in December.
What to agree on
BEREC draft documents state clearly that users must be able to decide the content they want to receive
or to send within their contracted Internet service not the ISP, neither Internet service can degrade
below a minimum service level; the threshold has to be determined by specific measurement
techniques, once set, it must be disclosed by ISP to NRA (National Regulatory Agency). There is a clear
distinction between the concepts QoS (Quality of Service), QoE (Quality of Experience) and NP (Network
Performance) and it is concluded that the better approach to ensure a minimum level of service is to
determine KPI (Key Performance Indicators) on NP. Related to QoS I agree that its applicability must
ensure transparency.
FCC rules recognize that there is a difference between mobile and fixed ISP and therefore rules should
be considered separately, it is important for the FCC that the end user has total control on the contents
he/she receives within a legal framework, also the traffic management techniques must be transparent
and to conform to standard best practices. It is remarkable that the rules explicitly prohibit any type of
discrimination that is considered anticompetitive and set some guidelines to identify what is considered
as unreasonable discrimination. Related to QoS the rules allow different classes of QoS to be applied in
the network as long as they can be equally provisioned to all applications and traffic flows.
Prof. Van Schewick states that NN (Network Neutrality) regulation must preserve the open nature of the
Internet that allows innovation without permission and free cultural exchange, which in turn also allows
low costs of innovation for developers, but the regulation must be clear enough to provide an easy way
2. to determine which are punishable behaviors and which aren’t without incurring on high costs for
governments or NRAs, for example, by having to study several cases separately (case by case scenario).
Any measure taken by ISP to single out an application or class of application for differential treatment
isn’t beneficial for the market and interferes with user decisions about how to use the network. Related
to QoS, the professor considers that exclusivity of QoS provision to one application or more applications
within a class of “like” applications (applications sharing the same characteristics, like network protocols
or usage) should be prohibited, and that applicability of QoS must not be dependent on which
applications users run on the network. Charging for any differential service provided by the network
should only be allowed to ISP own users. And finally NRA is responsible for the vigilance and compliance
of minimum level of service.
Where My Point of View Diverges
Image courtesy of Paola Buelvas (papolareina@yahoo.es)
Findings within the BEREC framework are very loose; the need to perform a case by case study of all
presumable infringements of the regulation is inferred, this leads to high costs for governments and
NRAs. Results are applicable to both fixed and wireless ISP, but the proposal to measure a mobile
operator performance is based on a periodical execution to produce geographical and accumulated
results… that means more costs!
FCC rules for mobile/wireless ISP give them permission for a reasonable traffic management but there is
no way to determine what is reasonable and what is not, this also leads to high costs and as commented
above its not beneficial for the industry.
According to Prof. Van Schewick the network must be application blind because this condition has
enabled the development of the Internet in the past and users can get exactly the QoS that meet their
preferences in a network that is completely blind to applications. The control of network resources can
be completely perform by using application agnostic network management and user control
prioritization, and it is possible to prevent aggressive users from overwhelming the network by
allocating bandwidth among users in application-agnostic ways. The concept of “like” applications is not
well defined, and network providers have broad discretion to decide which applications are alike, also
“like” treatment violates the principle of user choice.
3. I’m willing to Support My Views
The increasing usage of mobile internet along the rapid adoption of 4G across the world is an important
factor to include wireless/mobile ISP in the network neutrality regulation debate, according to Cisco VNI
the number of mobile-connected devices will exceed the world’s population by 2012. The main issue is
that wireless ISP are bounded by a big restriction called government regulated electromagnetic
spectrum usage, (at least for those interested in serving through the regulated portion of the spectrum),
while for a fixed ISP in order to expand its capacity it has almost complete autonomy across the
network, it cannot be said the same for wireless providers, so for them to expand capacity, government
auctions have to be conducted, accompanied by a thorough frequency planning and continuous
optimization. The other issue is the network blindness convenience; an intelligent network, aware of the
applications running on top, can better protect user personal information, it can help improve user
experience through complicated statistics gathering, it can provide exactly the type of quality of service
that the user wants for the application he wants how he wants it, it can help innovate through
collaboration with Internet Content Providers (ICP) sharing information and analytics about usage
patterns. Since the nature of mobile broadband usage is dynamic, subject to the environment, to the
amount of users and their positioning in the cell, to the technology employed by the ISP, to the type of
terminal running user applications and to other factors, the congestion control rarely can be effective by
performing fixed bandwidth allocation let alone using only application-agnostic ways.
The concept of “like” applications although can be viewed as not entirely well defined, it cannot be
defined as totally misleading; it is hard to believe that an ISP can convince a NRA that for example a P2P
application behaves like a VoIP application, sure, there might be occasions where an application can
behave like many, but Internet has a standard stack of protocols. On the other hand the statement that
an ISP that treats equally applications behaving alike violates the user choice is only true if the user
cannot choose otherwise, there are savvy users that will benefit greatly from “like” QoS treatments, and
ISP must be ready to offer that choice. Finally, transparency must be the key in the regulation starting
from the establishment of the baseline level of the service, it cannot be expensive or complicated so it
can be used globally as a guideline. The IEEE paper from Doctor Scott Jordan gives a technical deep
sustentation within a regulation framework.
Wireless QoS offering proposal within a NN framework
Considering all the factors presented in this blog, my proposal for a wireless ISP to provide and provision
QoS includes thoughs for the ISP the ICP (Internet Content Provider) and the users.
The proposal complies with the following principles:
1. User choice about his usage of the network
2. Innovation without permission both for ICP and wireless/mobile ISP
3. Conformity with standardized best practices of the industry
The proposal leaves out the following principle:
Network application blindness
QoS offering:
The Wireless/Mobile ISP must:
4. Let users decide how to apply QoS for its Internet service:The user will always decide how his
Internet usage will be prioritized, how his applications will be treated and how his subscription
will be handled by the network in case of congestion situations.
Provide different QoS classes:These classes of QoS will not be exclusive for applications or type
of applications, but can be applicable to either one based on user selection or preference. Under
this scenario all users, applications, and classes of applications have access to the same network
resources in non-discriminatory manner, only the user can select who receives what.
Provide information/education for the user:In order for the user to select the QoS class that
fulfills his preferences in a better way, the ISP must provide updatable information on how its
QoS classes work to prioritize the behavior of applications or applications itself. This in turn
provides the ISP the opportunity to set itself apart from the competition.
Charge only its users for QoS differentiation:In order to avoid that ISP charges twice for the
same service (ICP and users) prioritizing the same behavior or application.
Specialized Services network architecture could be open to ICPs:Since there is always the issue
of ISP specialized services competing for network resources with Internet services, then
wireless/mobile ISP can provide the ICP with the same open architecture it uses for its
specialized services (Web content servers closer to the users, IMS, etc.) so ICP applications can
benefit from ISP privileged position in the service delivery chain. This network usage can be
charged to ICP, but will be available to all ICP without discrimination.
Network Congestion control must be fully disclosed:Since the user picks how his service must
work, the ISP must be transparent to the user on how the network will manage his subscription
(Golden, Silver, Bronze) and his networks flows (QoS Classes behavior) in the case of network
congestion so the user can chose among different types of subscriptions and QoS classes.
Content providers can:
Innovate without permission:ICP and developers are free to create new applications without the
need to ask permission or establish previous communications to ISPs
Inform its user for better QoS selection:The more information the user gets the better decisions
on QoS can be made.
Use ISP Network to improve service delivery:Optional for those ICP and developers that see any
advantage for using the ISP network to enhance the user experience.
Users should:
Chose the options of QoS and configure the service according to their needs and preferences
Be informed about the best way to configure their service and the type of QoS that better suits
their needs
Follow me, @jomaguo.
For more discussion on everything mobile, visit the Service Provider Mobility Community at
www.cisco.com/go/mobilitycommunity.