User Perception of Mobility in NGN
Pablo Vidales, Marcel Wältermann, Blazej Lewcio, Niklas Kirschnick, The Mobisense project investigates user perception of VoIP services in mobility
scenarios, and feeds back the results from this assessment to existing perception
Introduction Sebastian Möller models. Figure 4 depicts the project methodology.
A speech sample is coded and transmitted under experimental conditions using the
The Mobisense project at Deutsche Telekom Laboratories aims to investigate and Mobisense testbed. These conditions can be dynamically changed as the experiment
quantify user perception of mobility in Next Generation Networks (NGN). From two occurs, this has the goal to emulate mobility situations and user mobility
views, networking and usability, the Mobisense project evaluates perceptual patterns:
mobility phenomena related to a diversity of link layer conditions in heterogeneous •attachment point to the Internet can be switched (network handover),
wireless networks, and maps the results to parametric user perception models.
Mobisense employs extensive auditory tests to quantify VoIP user satisfaction •underlying network conditions can be manipulated (e.g. data loss simulation),
according to the investigated network conditions, and so enables to meet customer •applied codecs can be changed.
expectations during the design of mobile VoIP services . An experimental setup The Mobisense project combines networking and usability evaluation methods. On the
already exists, which enables emulation and evaluation of multiple mobility one hand, network traces that are collected during the experiments are analyzed in
settings for ubiquitous wireless networking. order to discover the impact of networking phenomena on voice transmission quality.
On the other hand, the speech samples are recorded and evaluated during extensible
auditory tests that investigate overall call quality (user perceived quality) but
also instantaneous quality fluctuations and their correlations with user memory
(“recency effect”) and other non-study phenomena. Finally, the results of both
Mobisense research scope evaluations are correlated and integrated into one perception model and mobility
management protocols, which consider mobility aspects in heterogeneous networks, as
Implementation of a testbed to emulate NGN scenarios. The Mobisense testbed well as, the quality benefits resulting from wideband-narrowband-wideband
emulates multiple communication and mobility scenarios in NGN. The experimental transitions.
setup enables mobility across homogeneous and heterogeneous wireless networks.
Analysis of vertical/horizontal handover impact on VoIP. Future VoIP services and
seamless network roaming across multiple access technologies should meet customer
quality expectations. Therefore, user perceived quality of network handovers should
be quantified and mapped to the underlying network conditions.
Analysis of wideband-narrowband-wideband voice quality transitions. New voice
compression methods and powerful processors set the trend to wideband-narrowband-
wideband voice quality adaptation. However, user perception of voice quality
transitions is not fully understood. Figure 1 presents the design of the Soft Codec
Changeover solution  implemented as part of the Mobisense project. The solution
realizes voice quality transitions of ongoing VoIP calls, while mobile clients roam
across heterogeneous networks. Figure 1. The Soft Codec
Changeover is designed and
implemented to enable wideband-
narrowband- wideband voice quality
transitions, and VoIP service
adaptation. This solution enables
the elimination of application
related voice transmission
interruptions during codec
changeovers . Figure 4. The Mobisense methodology combines the best practices from both:
networking and usability research.
Figure 5. Sample test scenario to analyze
VoIP user satisfaction in NGN.
Evaluation of VoIP in “open broadband sharing” (aka Extended Hotspots)
scenarios. Efforts to deploy a global-scale wireless Internet access infrastructure
based on privately-owned DSL connections via IEEE802.11* are no grassroots anymore
(www.btfon.com). The connection owners (or the ISP) may control the amount of
shared bandwidth and so the impact on ongoing VoIP or multimedia sessions of third-
party users. The evaluation of user perceived quality on services running on a
shared infrastructure is also in the scope of Mobisense. The mobile user equipped with a dual-
Mapping of networking phenomena. Two-sided, networking and usability, mode mobile device initiates a VoIP
investigation of mobility phenomena in NGN must be mapped to a perception model call at home, using his home wireless
that will guarantee the improvement of mobility management and service adaptation connection, with luxurious wideband
Mobisense experimental setup
solutions in NGN. voice quality. However, he wants to
leave the house and continue his
The Mobisense project evaluates the impact of terminal mobility on user perception current call. He moves away from the
in NGN. The speech samples are processed and experiments are performed in the WiFi to switch to
As a result, he loses connectivity and is forced access point. an other available
Mobisense experimental setup, specially designed and deployed to cover the project network, HSDPA. As the network handover occurs, the mobile device automatically
requirements. Figures 2 and 3 present the network and software component views, switches the applied codec to narrowband, and so adapts to the new link conditions
respectively. . Some open questions are: How does the user perceive the network handover and
speech codec changeover? Is it possible to improve user experience? Would the user
accept this quality?
Conclusions, project status and future
Successful mobility management and service adaptation can improve user perception
of mobility in NGN. Therefore, Mobisense investigates various mobility situations
and the related user satisfaction to extend current models that will allow the
selection and design of mobility and service adaptation measures to provide
The diversity of wireless transmission technologies and changing link
characteristics while roaming across heterogeneous networks become a new challenge
for quality of mobile services. To close the gap, the Mobisense project studies the
problem of perceived voice quality by mobile users in NGN. Mobisense will try to
answer fundamental questions: What are the possible benefits and drawbacks of
performing network handover? What are the possible advantages of performing
Figure 2. The network-centric view of the wideband-narrowband-wideband quality transitions? When would the user make use this
Mobisense testbed. features?
Currently, the Mobisense testbed has been deployed, the mobility scenarios have
Figure 2 presents the networking view of the Mobisense testbed. It allows been emulated and the input data for the subjective test (listening) has been
connectivity to six networks attached to the Internet, which represent different collected .
transmission technologies: two LAN networks (one home and one foreign network), two Future work includes extensive auditory tests and network traces analysis to allow
WLAN networks, a HSDPA network, and a Flarion network. The home LAN network is the design of a user experience prediction model. The voice quality assessment is a
directly connected to the Internet and is the fix point of attachment for the Home
first step to build mobility management and service adaptation solutions that will
Agent and Correspondent Node. The Mobile Node roams among all networks.
improve user perception of VoIP services in NGN.
 Alexander Raake, Speech Quality of VoIP, Wiley, 2006.
 Marcel Wältermann, Blazej Lewcio, Pablo Vidales, and Sebastian Möller, A
Technique for Seamless VoIP-Codec Switching in Next Generation Networks (submitted)
 B.Lewcio, Assessment and adaptation of voice quality in All-IP heterogeneous
networks . Deutsche Telekom Laboratories (unpublished), September 2007.
 Mobisense Project Deliverable 1 (D.1), Experimental setup and test cases for
the assessment of VoIP in NGN (available on request)
Figure 3. Software components in the
Mobisense testbed. Acknowledgments
The software components are deployed on the Mobile and Correspondent node, and We would like to thank Prof. Anja Feldman, Dr. Marcin Solarski, Dr. Alexander Raake
for the support and fruitful discussions. We also thank T-Labs management for
these need to communicate in order to perform experiments. These communication supporting this project.
flows are shown in Figure 3. The testbed can be controlled in a centralized manner
from the Mobile Node, and it is managed by a controlling script that synchronizes
all software components.
For further information
Please contact email@example.com. More information on this and related
projects can be obtained at www.telekom.de/laboratory. A PDF-version of this
poster is available, too: www.deutsche-telekom-laboratories.de/~vidales