1. Abstract: Mobile devices are currently used not only
for two-way voice/data/video communication, but also
as wireless internet terminals. The power of
interactivity lies in the reciprocal process of
information exchange between two or more users and
an interactive service allows users to shop, vote, play
games, conduct financial transactions, reserve
holidays/ tickets and so on. This article highlights
importance and challenges of incorporating
“interactivity” in mobile multimedia system. Then,
various solutions (e.g., Hybrid-ARQ in Node B,
shorter transmit-time-interval, Node B-controlled
scheduling and adaptive modulation-and-coding) for
implementing interactivity are explained. Finally,
Node B-introduced ARQ and scheduling are examined
from a mobile system-standpoint to understand how
the solutions provide fast interactivity.
1 Importance of “Interactivity”
Mobile devices have allowed people across globe to
have two-way voice/data/video communication in a
tetherless manner. Use of the internet through mobile
terminals is also increasing rapidly throughout the
world. Mobile phones have evolved into wireless
internet terminals, enriching people's lives by letting
them acquire any type of information, whenever they
want. An interactive service allows users to shop, vote,
bet, play games, conduct financial transactions,
send/receive emails, book holidays/ tickets and so on
and so forth. Thus, “interactivity” plays a key role in
people-to-device dialog and makes a user active
participant in the process.
The power of interactivity lies in the reciprocal process
of information exchange between two or more
"players" in communication. "Players" can be pupils,
facilitators, peers but can also be automated learning
resources and other databases. Order-entry and many
other business applications need interactivity in a
limited way whereas mobile gaming requires
significant interactivity. In general, faster response and
interactivity provide a great deal of advantages through
(i) quick-and-better decision-making, (ii) easy
learning, (iii) on-demand request-servicing and (iv)
entertainment.
In Section 2, we outline challenges of implementing
“interactivity” feature. In Section 3, we discuss various
solutions to support “interactivity” in mobile
multimedia system. In Section 4, we explain how these
solutions facilitate providing “interactivity” feature in
next-generation mobile system. Section 5 concludes
the article.
2 Challenges of Incorporating “Interactivity”
As the applications using interactivity become more
demanding in terms of data rate and response-time,
there is an increasing need to improve system capacity,
enhance throughput and reduce overall communication
delay. At the same time, specific application and user
expectation are also important attributes to decide
whether the available data rate and responsiveness are
sufficient for provisioning interactivity. The meaning
of “interactivity” for a mobile user accessing
transactional information using SMS will definitely be
very different than a user playing network-based
mobile game.
In a typical 3G mobile system, radio access layer (as
specified in 3GPP Release 4) enables interactivity.
Fig.1 shows how communication takes place between
radio-link control (RLC) layer of user equipment (UE)
and RLC in the radio network controller (RNC)-side.
Large round-trip delay due to RLC retransmissions of
erroneously received data makes the system slow. This
does not support the delay-sensitive applications even
the data rate is not a constraint.
‘Interactivity’ in Mobile Multimedia Systems
Manoj Kumar(manoj-dsp.kumar@st.com), Alok Mehta, Kapil Soni and Aloknath De,
STMicroelectronics India

2. Figure 1: Radio Interface Protocol (3GPP Release 4)
Thus, a continuous need for even better spectral
efficiencies, reduced round-trip delay, improved user
experiences and new services is driving the
standardization of new features in mobile multimedia
systems.
3 Solutions to Support Interactivity
In order to meet the challenges discussed in Section 2,
the existing technology needs to be evolved for
increasing the system capacity and reducing the round-
trip delay. Increased system capacity will help to
provide higher data-rates and/or serve more number of
users. A large (e.g., 10 ms) Transmission Time
Interval (TTI refers to time-interval at which higher
layer provides data blocks to physical layer) and Radio
Link Control (RLC) layer-embedded ARQ protocol
makes the round-trip delay large in mobile systems
(ref: 3GPP Release 4). Hence, to reduce the round-trip
delay, a shorter TTI and an ARQ mechanism at the
NodeB level (Hybrid-ARQ) have been introduced.
This increases the system throughput also. In addition
to the above techniques, NodeB-controlled scheduling
of users and adaptive modulation-and-coding (AMC)
have also been introduced. These help in handling the
bursty traffic efficiently and enhancing the system
capacity/throughput. To implement Hybrid-ARQ and
NodeB-controlled scheduling, MAC functionality is
required to be placed partially at NodeB as shown in
Fig.2. Details of the schemes are given in the
following sub-sections.
Figure 2: Radio Interface Protocol (ref: 3GPP)
3.1 Hybrid ARQ in NodeB
A mix of forward error-correction (FEC) and
automatic repeat request (ARQ) is used to get reliable
and efficient communication over noisy channels.
ARQ ensures very low undetected error probability by
increasing the number of re-transmissions. On the
other hand, FEC corrects any error introduced due to
noisy channel and thus reduces the number of re-
transmissions. NodeB Hybrid-ARQ (HARQ) allows
rapid re-transmissions of erroneously received data
units. Since re-transmissions happen from NodeB (and
not every time from RNC), it reduces the number of
RLC retransmissions and the associated delays. This
definitely improves the quality-of-service (QoS)
experienced by the end user. As a NodeB-controlled
re-transmission is less costly from a delay perspective,
the physical channel can be operated with somewhat
higher error probability than in the system not
employing NodeB Hybrid-ARQ. This results in
improved system capacity. Soft combining can further
improve the performance of a Node B Hybrid ARQ
mechanism.
3.2 Reduction of Minimum TTI
Size of frame at the air interface, which equals the
minimum TTI, directly adds to the round-trip delay.
Thus, reducing the minimum TTI from 10ms to a
lower value (say, 2 ms, an optimal value considering
tradeoff between frame overheads and delay) will
reduce the transfer delay due to reduction of air-
interface delay and delay due to TTI alignment
(incoming data to be transmitted has to wait until the
start of the next TTI). A shorter TTI also helps to
reduce processing time of payload and round-trip delay
in Node B Hybrid-ARQ protocols, thus, resulting in a
higher system throughput and better resource
utilization.
RLC
UE NodeB RNC
IubUu
PHY
RLC
MAC
PHY TNL TNL
MAC
Shorter Round
trip delay path for
Hybrid ARQ
Partial
MAC
UE NodeB RNC
IubUu
PHY
RLC
MAC
PHY TNL TNL
MAC
RLC
Round trip delay
path for RLC
based ARQ
RNC – Radio Network Controller
RLC – Radio Link Control layer
MAC – Medium Access Control layer
PHY – Physical layer
TNL – Transport Network layer
Uu – WCDMA Radio Interface
Iub – Interface between NodeB and RNC

3. 3.3 NodeB-Controlled Scheduling
For distributing the system capacity and the available
power among all users in an optimal way, resource
scheduling needs to be done properly. Traditionally,
user scheduling is done by RNC on the basis of
channel conditions. This causes slow adaptation to the
changed channel conditions. Now, NodeB is given the
scheduling capability which ensures faster adaptation
to the channel conditions of each user.
In Fig.3., UE maintains a table containing all possible
Transport Format Combinations (TFCs) configured by
RNC. TFC quantifies the data which needs to be
processed by UE in a single TTI. This complete set of
TFCs is called Transport Format Combination Set
(TFCS). Traditionally, UE chooses the TFC from this
whole TFCS. However in NodeB-controlled
scheduling, a subset is defined by NodeB within TFCS
so that UE is restricted to choose TFC from NodeB-
defined TFC subset. NodeB can further update the
subset as per the interference and user buffer
information. This tighter control of uplink interference
results in increased capacity and improved coverage.
Node B controlled
TFC subset
TFCS configured
by RNC
Minimum SetTFC
TFC
TFC
TFC
TFC
TFC
TFC
TFC
TFC
TFC
Figure 3: Configuration of TFCs table at UE
In the downlink, the users are scheduled using the
various packet scheduling algorithms (based on
Channel Quality) which are as follows:
- Round Robin: The users are being served in a
cyclic order ignoring the channel quality
conditions.
- Fair Throughput: It provides a same
throughput distribution among the users.
- C/I Based: Scheduling strategies based on C/I
policy favour users with the best channel
conditions.
3.4 Adaptive Modulation and Coding (AMC)
The benefits of adapting the transmission parameters
to the changing channel conditions of a wireless
system are well-known. In fact, fast power control is
an example of a technique implemented to enable
reliable communications, while simultaneously
improving system capacity. The process of modifying
the transmission parameters to compensate for the
variations in channel conditions is known as link
adaptation. Another technique that falls under this
category is adaptive modulation and coding (AMC).
The principle of AMC is to change the modulation and
coding format in accordance with variations in the
channel conditions, subject to system restrictions. The
channel conditions can be estimated e.g. based on
feedback from the receiver. In a system with AMC,
users in favorable positions, e.g., users close to the cell
site are typically assigned higher-order modulation
with higher code-rates (e.g., 64-QAM with R=3/4
Turbo Codes), while users in unfavorable positions
e.g., users close to the cell boundary, are assigned
lower-order modulation with lower code-rates (e.g.,
QPSK with R=1/2 Turbo Codes). The main benefits
of AMC are: a) availability of higher data-rates for
users in favorable positions which in turn increases the
average throughput of the cell, and b) reduced
interference variation due to link adaptation based on
variations in the modulation/coding scheme instead of
variations in transmit power.
4 System Analysis
Let us consider the example of world-wide-web access
through mobile phones. We not only interact with the
web browser, but also with the pages that the browser
brings to us. The implicit invitations called hypertext
that link us to other pages provide the most common
form of interactivity when using the web. This can
then be thought of as a giant, interconnected
application program. Let us now investigate the impact
of techniques on the complete UMTS system. As our
focus is on NodeB sub-system, discussion will explore
the system aspect at NodeB level primarily.
4.1 Intelligent Node-B and System Complexity
By embedding intelligence in NodeB, it has been given
the capability to detect error(s) in received data frames
and accordingly take action (sending ACK/NACK to
UE for uplink). It also receives the ACK/NACK
corresponding to the frames sent to the UE in

4. downlink and retransmits the frames erroneously
received by UE. Scheduling functionality has also
been partially shifted from RNC-MAC to NodeB.
These added functionalities make NodeB more
complex and require extra amount of processing
power. As the round-trip delay also constitutes the
NodeB decoding time, it is imperative to reduce the
processing time. For implementation, a more complex
NodeB requires high processing CPUs and DSPs,
sufficient on-chip memory, and fast off-chip memory.
4.2 Performance of RNC Vs NodeB Scheduling
In centralized scheduling, the scheduler is located in
RNC, and is responsible for simultaneous scheduling
of UEs across multiple cells. Thus, it is possible to
take into account the impact of each scheduled UE in
all cells of its active set. However, the drawback of
such scheme is the significant scheduling delay. To
reduce the scheduling delays and take advantage of the
possible fast scheduling gains, NodeB scheduling is
needed. In this case, the scheduling is decentralized as
the knowledge of the received signal level is available
only at the scheduling NodeB and each NodeB
schedules the UEs without considering their
contribution to the other cells. Hence, there is an
advantage of the decentralized scheduling over the
centralized scheduling due to the shorter delays
incurred in the scheduling process and the possibility
of exploiting the fast scheduling gain. However, the
lack of knowledge of the impact an UE may have on
the other cells’ rise-over-thermal noise (RoT) is a
disadvantage.
The objective is to determine the loss due to the partial
information availability in the decentralized scheduler,
without exploiting any possible fast scheduling gains,
and form a benchmark for the gains needed to be
provided by faster scheduling. Based on the simulation
results it can be seen that for the same average RoT,
the centralized scheduler yields a throughput gain over
the decentralized scheduler. Also, while the fairness
remains the same, the RoT overshoot (Probability
{RoT > 7dB}) is higher in the case of decentralized
scheduling. This degradation represents the minimum
benchmark for the gains to be provided by faster
scheduling and other techniques that rely on faster
scheduling.
4.3 HARQ Efficiency Vs Block Error Rate
(BLER) operation point
To evaluate the efficiency of HARQ the BLER for the
first transmission should be a non zero value. HARQ
efficiency increases as the BLER is increased.
However from complexity point of view, the increase
of BLER operation point (to 50% or even higher) also
means complexity increase. If the first transmission
BLER is high and the single user throughput is kept
the same, then the peak data rate has to be increased.
This signifies allocation of more hardware and/or
software resources for each user both at UE and radio
network side. On the other hand, if single user
throughput is allowed to decrease and the system
throughput is increased by allowing more users in the
cell, it requires more hardware and/or software
resources in the network side.
The amount of Iub traffic and buffering and processing
at RNC is increased if the reordering is performed at
RNC. Roughly speaking, BLER=50-60% operation
point requires that the baseband buffers and amount of
processing resources are doubled compared to
BLER=10% operation point. The higher peak data rate
per user or more users both imply that the amount of
processing and the amount of buffering need to be
increased at Node B.
5 Concluding Remarks
Order-entry and other business applications need
interactivity in a limited way, whereas video
streaming, mobile gaming, virtual office, tele-learning
and similar applications require significant
interactivity. In general, faster response and
interactivity provide a great deal of advantages through
quick-and-better decision-making, easy learning, on-
demand request-servicing and entertainment. In this
article, various solutions (e.g., Hybrid-ARQ in Node
B, shorter transmit-time-interval, Node B-controlled
scheduling and adaptive modulation-and-coding) for
implementing interactivity have been explained
independently as well as with reference to a mobile
system. Pens, thumb/phone keyboards, touch screens,
etc. may be appended to mobile multimedia systems so
as to provide user interface for proper interactivity.
REFERENCES
[1] 3GPP TR 25.896 v6.0.0 (2004-03), Feasibility
Study for Enhanced Uplink for UTRA FDD
[2] 3GPP TR 25.848 v4.0.0 (2001-03), Physical Layer
Aspects of UTRA High Speed Downlink Packet
Access