CONVERGED OPTICAL NETWORKS FOR VIDEO AND DATA

DISTRIBUTION IN HOSPITALITY ENVIRONMENTS

Iñigo Artundo, David García-Roger, Beatriz Ortega and Jose Capmany

Optical and Quantum Communications Group, iTEAM, Universidad Politécnica de Valencia (UPV)

Camino de Vera s/n, 46022 Valencia, Spain

Keywords: Optical Networks, In-building Hospitality Networks, IPTV, Plastic Optical Fiber, Radio-over-Fiber.

Abstract: Current hospitality networks are already lagging behind in terms of broadband adoption and high-speed

online offered services, and they might not be able to cope with the increasing bandwidth demands required

to distribute high-definition video traffic. We propose the use of a converged optical network adapted to the

specific needs of the hospitality environment, providing a low-cost and low-power solution based on a

combination of silica and plastic optical fiber wiring, together with radio-over-fiber techniques for wireless

access. We use detailed full-system simulations to analyze the validity of such infrastructure to provide a

unified, pervasive and future-proof all-optical solution.

1 INTRODUCTION

The recent growth of demand for higher bandwidth

in hospitality buildings (hotels, hospital, residences,

etc.) comes from the exploding connectivity of

devices and systems sharing data, the high-tech

customization of guestroom technology (video

streaming to TV and other devices, on-demand

movies, high-speed Internet access, online gaming,

voice-over-IP and videoconferencing, e-commerce

and billing of the resort's services), and the

increasingly pervasive mobile access to information

(Cisco, 2009). Moreover, remote guest/staff control

of room temperature and lighting (including energy

management policies), electronic door locks, or

other hotel services will only add to the number of

connected and managed devices. Emerging

technologies will strongly affect network

performance and medium/long-term strategies to

meet guest demands while still providing consistent

quality of service and experience (Inge, 2009).

The deployment of mixed physical technologies

on the network infrastructure (coaxial, Cat.5e/6,

phone wire) results in large capital equipment

installation and maintenance costs. Moreover, the

common electrical cabling used poses limitations in

diameter, weight, terminations and

electromagnetical interference. One growing

approach to simplify network design, installation

and management costs is the use of converged

networks, both wired and wireless, using a single

cable for backbone and in-room wiring, to carry all

communications for guestrooms as well as for staff

systems. Here, the larger bandwidth of fiber over

copper links becomes evident in high-density

hospitality buildings, where user aggregation

reduces installation and operation costs; moreover,

recent developments on Plastic Optical Fiber (POF)

(Nespola, 2010) and Radio-over-Fiber (RoF)

transmission allow for low-cost and competitive

converged all-optical architectures (Gomes, 2009).

In this paper, we study future requirements in

hospitality environments to introduce high-definition

video services through converged optical networks.

In Section 2 we identify the design boundaries of

such scenario through a statistical analysis of TV

channel blocking and Section 3 proposes a future-

proof fiber-based optical architecture that provides

wired and wireless services, and the limitations on

the span of such network are thoroughly described.

Finally, Section 4 introduces full-system simulations

of this architecture to analyze network response over

intensive video traffic, depending on the floor

distribution capacity. This allows the exploration of

the concept of flexible optical resource allocation to

cope with temporal limitations and cover specific

hospitality services in a dynamic way.

139

Artundo I., García-Roger D., Ortega B. and Capmany J. (2010).

CONVERGED OPTICAL NETWORKS FOR VIDEO AND DATA DISTRIBUTION IN HOSPITALITY ENVIRONMENTS.

In Proceedings of the International Conference on Data Communication Networking and Optical Communication Systems, pages 139-145

DOI: 10.5220/0002988601390145

 SciTePress

Table 1: Bandwidth requirements to support hospitality HD IPTV.

External WAN

Maximum

bitrate (Mb/s)

Average Throughput

HDTV channels

(6 Mb/s/ch)

Full HDTV channels

(12 Mb/s/ch)

(Mb/s) %

ADSL 8.2 8* 97.5% 1 0

ADSL2+ 24.6 21.8* 88.6% 2 1

VDSL2 250 68* 27.2% 11 5

Gigabit Ethernet/GEPON 1000 941.5 94.1% 156 78

GPON 2500 2488 99.5% 414 207

10Gigabit GPON/ 10GEPON 10000 9415 94.1% 1569 784

In-building LAN

Fast Ethernet 100 94.2 94.1% 16 7

Gigabit Ethernet 1000 941.5 94.1% 156 78

WiFi 802.11g 54 ~10-22 18-40% 1-3 0-1

WiFi 802.11n (2.4/5 Ghz)

300/

600

~65-260/

~135-540

21-86%/

22-90%

10-43/

22-90

5-21/

11-45

* Obtained from empirical measurements over 2500 m. (Kagklis, 2005). VDSL2 average is calculated theoretically for this same distance.

2 PREVIOUS WORKS

Previous works have attempted to measure the

impact of technological investment in hotels (Sigala,

2004), reaching the conclusion that there is a

significant productivity impact when the exploitation

of the network and its integration with the

infrastructure are strategically optimized, with

architectures adapted to the business and its

operations.

Similar converged optical architectures have

been proposed recently but mostly for generic in-

building scenarios. For example, (Xu, 2010)

examines the transmission of uncompressed High-

Definition TV (HDTV) under Ethernet passive

optical networks (EPON) making use of the

polarization diversity technique to improve reception

sensitivity and increase the anti-interference capacity

of the in-building wireless transmissions. (Walker,

2009) showcases a 600 Mb/s radio-over-fiber

architecture able to integrate simultaneously Ultra-

Wide-Band (UWB), wireless and WiMax signals

over 1 km of silica fibre using reflective electro-

absorption transceivers, and similarly, (Jia, 2009)

proposes multi-band generation and transmission of

all these wireless signals through photonic frequency

tripling, demonstrating a testbed able to deliver

uncompressed 1.485 Gb/s HDTV video signals over

silica fiber and air links.

Regarding the use of plastic fiber for RoF

transmission, it has been already demonstrated in

(Guillory, 2009) on a home level, with a network

delivering various signals (Ethernet, digital

terrestrial or satellite TV, wireless) on graded-index

POF, and very high bit rates - up to 16 Gb/s - have

been achieved by using techniques such as

orthogonal frequency-division multiplexing (Yu,

2008).

However, none of the architectures reviewed to

date addresses the specifics of the hospitality

networks, nor tries to provide a complete optical

solution for the complex structured cabling systems

commonly used in these commercial infrastructures.

3 FUTURE VIDEO BANDWIDTH

REQUIREMENTS

In hospitality installations, IP television (IPTV)

networks are quickly substituting traditional RF

video systems due to its lower installation and

maintenance costs, and the possibility to integrate

other interactive hospitality services (Held, 2006).

IPTV systems use a single type of data connection

for television, data and telephony services,

benefiting from the use of a single set of cabling.

They commonly receive TV channels from either

cable television or satellite connections, and they

may also introduce receiver antennas to get free

local broadcast channels. A key feature for the

upgrading of hospitality TV systems to have IPTV

capability is the use of a hybrid solution that can

combine simulcast television content (analogue and

digital distribution) with multicast and unicast IPTV

channels.

However, high-definition IPTV distribution faces

specific challenges in the hospitality scenario. The

number of channels that must be available to guests

tends to be higher than in residential deployments, as

overcoming geographic boundaries is critical for

DCNET 2010 - International Conference on Data Communication Networking

140

travelers. Video-on-Demand (VoD) is highly

promoted, as it offers relatively high margins

(a)

(b)

Figure 1: Channel blocking probability for (a) K = 200

HD-MPEG4 channels, and (b) K = 44 large bitrate (e.g.

3DTV) channels.

and is a large part of the hospitality TV service

revenues, but it puts an additional demand on

temporal extra bandwidth requests. In Table 1,

bandwidth demands for the distribution of high-

definition IPTV channels are shown, not considering

any additional type of traffic. Channel encoding is

considered to be MPEG4/H.264 HiP (1280x720 or

1920x1080) at 24 frames per second. It is obvious

that there is a clear limitation when using current

hospitality deployments, consisting of different

Digital Subscriber Lines (xDSL) to the building and

Fast Ethernet or wireless in-door access, and will

even become worse when extra video streams like

security cameras or higher bitrate video services,

like 3D TV (up to 24 Mb/s) are also included.

To analyze the designs boundaries of a common

hospitality video distribution network, we will first

focus on calculating the blocking probabilities for

each TV channel under a xDSL-driven IPTV

installation. Using the models proposed by (Lu,

2007), the blocking probability B(i) for channel i

will be given by Equation 1, with B

Proc

the blocking

probability due to limited processing capability of a

Digital Subscriber Line Access Multiplexer

(DSLAM), and B

Link

(i) the blocking probability due

to insufficient link capacity from a DSLAM to an

edge router:

() (1 ) ()

Proc Proc Link

iB B Bi=+− ⋅

(1)

(1)!

(1)!!

(1)!

(1)!!

DSLAM

Proc

DSLAM

snn

shh

−

⋅

−−

−

⋅

−−

∑

(2)

() (1 ()) ()

Link Engset

iPiBi=− ⋅

(3)

In these equations, s corresponds to the number of

users per DSLAM, n is the channel replication

capabilities of the DSLAM, and ρ

DSLAM

is the ratio

between guest arrivals and departures to the

DSLAM, according to a Poisson distribution.

() 1

() ()

Engset Engset

eB ieB i

ρρ

=−

−⋅+

(4)

()

Engset

qpz

mdz

qpz

jdz

⎡⎤

⎢⎥

⎣⎦

⋅

⎡⎤

⎢⎥

⎣⎦

∏

∑

(5)

P(i) represents the probability of channel i to be

‘on’, as shown in Equation 4, with ρ

as the ratio

between guest arrivals and departures to channel i.

Engset

(i) described in Equation (5) is the probability

that the link from the DSLAM to the edge router is

consumed by m channels other than the requested

channel i. If we consider C as the link capacity

coming into the building, and C

the channel bitrate

(depending on the compression coding), we can

calculate how many m = C/C

of the K offered

channels can be transmitted simultaneously to N

guests using M DSLAMs in peak usage hours, which

normally correspond to a 41% guests connected

between 16:00 and 22:00 (SKO, 2008). We will

consider a single DSLAM for 180 guests. Channel

switching rates are distributed with decreasing

exponential popularity, according to a realistic

distribution (SKO, 2008).

We will consider a non-blocking situation when

B(i) < 10

-8

for all popular channels. Video bitrates

depend on the encoding, 3 Mb/s for SDTV using

MPEG2 or 1.6 Mb/s with MPEG4, and 12 Mb/s and

6 Mb/s accordingly for HDTV. In this case, the

minimum non-blocking number of channel

replications per DSLAM would be n = 65, although

we will consider a more realistic value of n = 120.

Table 2 shows the maximum available channels and

CONVERGED OPTICAL NETWORKS FOR VIDEO AND DATA DISTRIBUTION IN HOSPITALITY

ENVIRONMENTS

141

(a) (b)

Figure 2: (a) Converged optical network architecture, and (b) building elevation and floor plans.

Table 2: Maximum available channels and minimum

incoming link capacity.

TV Channel

bitrate (C

)

Max

for

C=1000 Mb/s

Min

for K=500 ch

1.6 Mb/s 674 ch 750 Mb/s

3 Mb/s 357 ch 1410 Mb/s

6 Mb/s 174 ch 2830 Mb/s

12 Mb/s 85 ch 5560 Mb/s

minimum incoming link capacity for different IPTV

channel bitrates.

If we set the link capacity to C = 1 Gb/s, there is

a clear limitation on the number of non-blocking

HDTV channels that can be available. Considering a

future offer in the order of 500 channels, the link

capacity quickly scales up over the gigabit range due

to fragmented channel viewing, limiting the use of

traditional coax or twisted pair approaches. Figure 1

shows (a) the channel blocking probability for the

same scenario in the case of 200 HDTV channels at

6 Mb/s bitrate, and (b) a future scenario with much

larger bitrate channels (e.g. 3DTV) at 24 Mb/s

bitrate. It can be clearly seen that only the two most

popular channels are absolutely blocking free, and as

channel popularity decreases (higher channel index),

blocking can raise even four orders of magnitude.

For a higher bitrate, blocking due to channel

capacity can happen even for the most popular

channels. When reducing n to 60 processing

limitations quickly appear at the DSLAM for all

channels alike.

If we now consider a Gigabit Passive Optical

Network (GPON) link capacity of C = 2.5 Gb/s,

when using HD-MPEG4 encoding we would be able

to access 446 channels in a non-blocking way for

this same scenario. In case of having wireless IPTV

distribution, with links up to C = 600 Mb/s, we

could transmit blocking-free only 103 channels. Of

course, the number of channels will be reduced in a

real implementation depending on bandwidth

reservations for management and other extra

services.

4 CONVERGED OPTICAL

CABLING

The proposed optical architecture, shown in Figure

2a, consists of a router that connects to the external

service/content providers through and optical access

network, either Fiber-to-the-Building GPON or

EPON connection. The building backbone from the

basement router to the different floors is done

through single mode fiber (SMF), and floor

distribution is done through step-index plastic

optical fiber (SI-POF), as shown in Figure 2a.

Wireless access will be provided by a Distributed

Antenna System (DAS), consisting of multiple

Remote Antenna Units (RAUs) capable of

transporting frequencies from 800 to 2500 MHz,

including 2.5G/3G mobile and 802.11/16 wireless

signals, through the SMF backbone.

To study the feasibility of such optical

infrastructure, a transmission analysis is done first.

Maximum floor link lengths will be mainly limited

by the high transmission loss of SI-POF. The design

boundaries then will be given by the transmission

losses for a generic building geometry, as shown in

Figure 2b. Here, M corresponds to the number of

floors with rooms, N the number of rooms per

corridor side, C the number of corridor sides, H the

floor height, L the room length, W the room width,

and L

the initial length from the floor switch to the

first room (considered to be always 4 m). In each

floor, a double SI-POF cable will be installed to

DCNET 2010 - International Conference on Data Communication Networking

142

Figure 3: POF transmission losses depending on the room length, width and number of rooms per corridor.

each room to provide two wall plugs (for data access

and an IPTV device). The basic floor distribution

includes a central hallway with connections to

security video-cameras and RAUs.

The transmission losses for SI-POF are

considered to be L

= 0.16 dB/m at 650 nm, and

0.09 dB/m at 510 nm. High-quality POF connectors

(EM-RJ, SMI) introduce a loss L

= 0.75 dB and

tight bends (around 5 mm radius) can introduce

losses of L

= 1.5 dB per bend. Therefore, the

maximum distance for any floor link will be given

by Equation 6, the minimum transmission losses for

the longest POF link calculated by Equation 7, and

the power margin for each POF link given by

Equation 8, with P

and P

the transmitter and

receiver power respectively, in dBm:

(6)

(7)

(8)

Considering a link with a standard Light Emitting

Diode (LED) with a transmitted power of P

= -3

dBm and a PIN photoreceiver with a sensitivity of

= -19 dBm, two connectors (1.5 dB) and six tight

bends (9 dB), the maximum transmission loss per

link would be 5.5 dB.

Figure 3 shows the link transmission loss for

different room configurations. Up to 30 m

room

sizes, the POF links will be able to span up to 5

rooms in each corridor side without significant

power loss. This would be enough to cover an

average hotel with 20 rooms per floor (N = 5, C = 4).

If using green LEDs, their lower attenuation would

allow extending the reach to more than 10 rooms per

corridor side. For the backbone RoF transmission, as

the bandwidth of SI-POF is limited to only 500

MHz·Km, either SMF or GI-POF with bandwidth of

2500 MHz·Km, would be needed. Maximum link

lengths in this case, considering full building span,

would be D

RoF_Max

= (M+3)·H, and considering the

same power budget, for an average hotel floor height

of 4 m, this would mean a maximum GI-POF link

span of 12 floors. For longer heights, silica fiber

links with lower losses would be mandatory for

keeping the power budget.

5 HOSPITALITY DATA TRAFFIC

To study future network demands of hospitality

environments under this converged optical

architecture, we have simulated the whole

infrastructure presented in Section 3 under

OPNET™ Modeler.

An average 4-floor hospitality resort is

considered, with 45 rooms per floor plus 45 staff

computers making a total of 225 users while in full

occupancy. This distribution mirrors a typical

medium size hotel or a small hospital. The main

building backbone is modeled to be 100 m SMF

links running Gigabit Ethernet, and floor distribution

is done through POF to wall plugs and through the

RAUs for the wireless access. Average POF link

lengths range from 6 to 50 m, and are considered to

run Gigabit Ethernet, although Fast Ethernet links

have been modeled as well for comparison.

Guest and staff users and devices have been

profiled daily in time following future application

usage distributions (ALPHA, 2008), and several

services have been included in the analysis as shown

in Table 3 and Figure 4. Application duration and

their repetition rate are modeled through uniform

and exponential distributions, and guest profiles are

modeled under a low (0.5), medium (1) and high (2)

usage factor over these values, with application and

guest start times modeled through the OPNET

Random Number Generators (PRNG), based on the

operating system's

random() implementation. We

⎥

⎦

⎥

⎢

⎣

⎢

++=

WNLD

MaxPOF

BCTMaxPOFMin

LLLDL ++= ·

RMinT

PLPPM −−=

CONVERGED OPTICAL NETWORKS FOR VIDEO AND DATA DISTRIBUTION IN HOSPITALITY

ENVIRONMENTS

143

performed daytime simulations to see the dynamic

evolution of traffic as users connect their devices

into the network, and to obtain peak bandwidth

demands, which will be dominated by the heavy

video traffic. As daily TV usage begins early in the

Figure 4: Guest profile definition.

Table 3: Application profile definition.

Application

(Mb/s)

Duration (s)

Repetition

(s)

Web browsing* 1-100 Exp(600) Exp(600)

Email 1 Exp(60) Exp(600)

Database access 0.5 Uni(300) Uni(300-

600)

VoIP 0.096 Exp(600) Exp(1800)

Videoconference 4 Exp(600) Exp(7200)

File transfer 100 Exp(300) Exp(1800)

IPTV 6-24 Exp(7200) Exp(21600)

* Includes also video browsing, social networking and immersive online

games and environments.

morning, we can see in Figure 5 the bandwidth

demands per floor, quickly over passing the Fast

Ethernet limit of 100 Mb/s. Making use of Gigabit

Ethernet seems mandatory to avoid congestion if

intensive video traffic (HDTV or high resolution

videoconferencing) is considered. This is specially

important in tele-healthcare applications. The

similarity among floors in the bandwidth traces over

the random statistical variability is due to the

identical guest profiling used per floor to simplify

design and simulation time and future work intends

to add further detail to such guest distribution.

As a way to measure the responsiveness of the

system, we we will use the channel switching time

of the hospitality IPTV system. The channel

switching time (or zapping delay) can be defined as

the time difference between the user asking for a

channel change by pressing a button on the remote

control and the display of the first frame of the

requested channel on the TV screen. In analog TV,

channel change is around 100 ms since it only

involves the receiver tuning to a new carrier

frequency, demodulating the analog signal and

displaying the picture on the screen. IPTV channel

switching times can be higher due to delay factors

(Uzunalioglu, 2009), like digital video

decompression and buffering, IP network related

issues (frame encapsulation, IGMP group joining,

congestion, etc.) and content management (paid

subscription channels, parental filtering, etc.). Fast

switching IPTV systems are expected to have

zapping delays of less than a second. To make our

study more general, we will only consider the

network components of the channel switching delay,

as video codification and content management

greatly depends on the specific IPTV

implementation.

In Figure 6 we see the influence of full HDTV

data streams (12 Mb/s per channel) over the channel

switch requests (the zapping delay). We observe

that, under a gigabit Ethernet network, this value

still falls far from the margin of 125 ms limit for

seamless zapping time, but over slower connections,

delay quickly builds up and can degrade user

experience and congest the remaining data network.

For the same situation, but considering all users

streaming high-definition WebTV video (6.3 Mb/s)

instead of IPTV, or high-definition

videoconferencing (4.3 Mb/s), there would be an

average frame delivery delay of 2.74 ms under a

Fast Ethernet network, or only 0.38 µs under a

Gigabit Ethernet one.

We can conclude that the presented architecture

remains valid for the services evaluated and still

holds enough margins to cope with spikes due to

seasonal trade show attendance or an increase in

holiday travelling. Moreover, dynamic network load

balancing on the optical domain can spread future

demands over all available physical media, avoiding

traffic surges, server bottlenecks, connectivity losses

and downtimes. Considering specially the RoF

distribution system, more wireless and mobile

capacity can be allocated to an area (e.g. conference

hall) during peak times and then re-allocated to other

areas when off-peak (e.g. guestrooms in the

evenings). This obviates the requirement for

allocating permanent capacity, which would be a

waste of resources in cases where traffic loads vary

frequently and by large margins. Future explorations

on the hospitality scenario will include the dynamic

use of different optical wavelengths through

reconfigurable Wavelength Division Multiplexing

(WDM).

DCNET 2010 - International Conference on Data Communication Networking

144

Figure 5: In-building data traffic at morning time.

Figure 6: (a) Channel zapping delay correlated to (b)

bandwidth consumption in a Fast Ethernet POF link.

6 CONCLUSIONS

We have mapped the requirements for high-

definition video distribution on hospitality

environments, identifying bandwidth demands of

more than 5 Gb/s. Thus, fiber emerges as the best

cabling solution, and thus we propose a low-cost and

future-proof converged optical network based on

silica and plastic fiber. Such cabling infrastructure

would be limited to 12 floors with 20 rooms each,

and full network simulations have verified TV

channel switching times of less than 125 ms.

ACKNOWLEDGEMENTS

This work was supported by the EC 7th Framework

Program: Architectures for fLexible Photonic Home

and Access networks (ALPHA), ICT CP-IP 212 352,

and OPNET Technologies, Inc. University Research

Program.

REFERENCES

Cisco Systems, 2009. Hyperconnectivity and the

Approaching Zettabyte Era, Whitepaper.

Inge J., 2009. Getting it Together: Technology trends in

the hospitality industry, Hospitality Upgrade

Magazine, pp. 14-26.

Nespola A. et Al., 2010. High-Speed Communications

over Polymer Optical Fibers for In-Building Cabling

and Home Networking, In IEEE Photonics Journal, to

be published.

Gomes N. J. et Al., 2009. Radio-over-fiber transport for

the support of wireless broadband services, In Journal

of Optical Networking, Vol. 8, pp. 156-178.

Sigala M. et Al., 2004. ICT Paradox Lost? Journal of

Travel Research, Vol. 43, pp. 180-192.

Xu K. et Al., 2010. Enabling RoF Technologies and

Integration Architectures for In-Building Optical–

Wireless Access Networks, IEEE Photonics Journal,

Vol. 2(2).

Walker S. et Al., 2009. Wireless Broadband Service

Delivery via Optical In-Home Converged Networks,

In Proc. of the IEEE Photonics Society Annual

Meeting.

Z. Jia, 2009. Wireless High-Definition Services over

Optical Fiber Access Networks, In Proc. of the 18th

international Conference on Wireless and Optical

Communications Conference, pp. 48-51.

Guillory J., 2009. Multiservice & multiformat home

network based on a low cost optical infrastructure, In

Proc. of the ECOC.

Yu J., 2008. 16Gbit/s Radio OFDM Signals over Graded-

Index Plastic Optical Fiber, In Proc. of the ECOC.

Held G., 2006. Understanding IPTV. Auerbach

Publications, Informa Telecoms & Media.

Kagklis D. et Al. 2005. A comparative Performance

Evaluation of the ADSL2+ and ADSL Technologies,

In GESTS International Transactions on Computer

Science and Engineering, Vol.19(1).

Lu Y. et Al., 2008. E2E Blocking Probability of IPTV and

P2PTV, Lecture Notes in Computer Science, Vol.

4982, pp. 445-456.

SKO Dutch Audience Research Foundation Annual report,

2008. http://www.kijkonderzoek.nl

ALPHA FP7 Project, 2008. End-user future services in

access, mobile and in-building networks, Public D1.1

deliverable.

Uzunalioglu H., 2009. Channel change delay in IPTV

systems, In Proc. of the 6th IEEE Conference on

Consumer Communications and Networking

Conference, pp. 206-211.

CONVERGED OPTICAL NETWORKS FOR VIDEO AND DATA DISTRIBUTION IN HOSPITALITY

ENVIRONMENTS

145