Page 25 - ITU-T Focus Group Digital Financial Services – Consumer Experience and Protection

ITU-T Focus Group Digital Financial Services
Consumer Experience and Protection

1.3 Is DFS a “Popular Service”?

DFS is popular, yes – but DFS is only a class denominator.

NOTE: At the time work on mobile QoS started (about 10 years ago), the experts considered “service” as
something which has a direct impact to the customer’s perception. Typical examples would be telephony
or web browsing. A “service” in this view is understood as something connected to an end to end use case.
However, many end to end use cases relate to “carrier services” (such as some type of packet data functionality
having their own QoS metrics (KPI).

In this context DFS can be considered as a classical example of such a user-related service, which can be realized
in several ways, using “carrier services” such as SMS or packet data functionality of networks.

DFS is not alone in this “top level service” view. Today's telephony is a prominent example. End users basically
do not care if the function they are looking for (being able to orally communicate with another) is realized
using legacy GSM or UMTS, VoLTE or some OTT VoIP technology. Their quality assessment is based on universal
metrics such as setup time, call drop rate or speech quality, which are exactly those metrics which are at the
core of standards such as Recommendation ITU-T E.804 or ETSI TS 102 250.
The sometimes very detailed KPI definitions in these standards are owed to a “diagnostic” approach, but by
no means not "the golden rule". Future developments will attempt to reveal true “end customer” related Key
Quality Indicators (KQI).
An additional example for this may be web browsing using HTTPS instead of HTTP. For the user, nothing seems
to have changed, so top-level QoS KPI to assess user perception are the same - however, the networks are
treating HTTPS and HTTP traffic in many cases differently, which will lead to a difference in usage of such KPI
for diagnostic purposes.

If we want to technically assess the expected top-level QoS of a particular DFS offering using a carrier service
point of view, we need to know the technical flow of data and signalization. This information is not normally
available from service providers’ websites or brochures.

NOTE: Strictly speaking this is true for most of the other services offered by network operators. First of all,
operators typically do not commit themselves (at least not towards end customers) to strict performance
targets; in the case of mobile networks this is perfectly understandable as the local conditions vary in a wide
range (e.g. from rooftop to cellar of a house even in the same geographical spot). Then, with networks going
even more towards “content sensitive” behaviour for the sake of resource optimization, the performance
cannot safely be predicted from just some general “bit pipe” properties, measured using simple end to end
services such as web browsing. However, DFS can be - as will be shown later - made subject to objective
measurement quite easily.

Ideally, this must be dealt with when licenses are negotiated between regulators and potential DFS service
providers.

NOTE: This is well known and understood for other services like for example, video streaming:

When “YouTube™” first became popular it was based on TCP streaming; with this information KPIs could be
defined in standards, QoS could be assessed and QoE could be predicted. Today, for good reasons, the same
service by the same entity is rendered as adaptive streaming using HTTPS. Consequently, new standards have
been written with new KPIs in order to assess QoS for the “same service”.

Strictly speaking, the KPI with respect to video quality are still the same; only the methods have changed (or
were forced to change). Most importantly, KPI definitions using “low level” technical events as those from the
IP level do not work anymore if encrypted connections such as HTTPS are used.

If we can identify categories of different DFS offerings, we could conclude, which of such categories constitute
“popular services” (i.e. which are widespread and used by many customers) and start a more selective look
into KPI definitions.

17