Technology White Paper
Media Servers (MS) will play a key role in future
IMS/NGN networks. More advanced services
require more and more complete interaction with the
user. Consequently, the Media Servers or Media
Resource Platforms that are deployed in today’s
networks need to be enhanced. IMS, first designed
for mobile networks, is now being promoted as a
converged solution for both fixed and mobile
networks. IMS-compliant Media Servers as defined by
IMS standards, aim to provide a decoupled and
flexible architecture, thanks to an optional split of the
Media Resource Function Control (MRFC) and the
Media Resource Function Processor (MRFP). The
objective of IMS-compliant media resource functions
is to allow media resources to be shared among
various applications to facilitate the rapid creation of
services.
This paper focuses on the evolution of Media Servers.
These have moved from stand-alone devices catering
to dedicated applications towards flexible devices
capable of voice/video/Web services integrated with
multimedia end devices. Furthermore, Media Servers
are poised to play a significant role in triple-play
services. They will serve requests from consumer’s
set-top boxes and mobile multimedia terminals/browsers
for additional video play-out and streaming. The
paper also looks at some advanced aspects of
managed customer interaction and the applications of
Media Servers in mobile networks. The required
protocols and extensions to existing protocols to
achieve such advanced services are addressed, too.
The paper concludes with a summary of the
positioning of Alcatel’s offering, including the Alcatel
8688 MRF and 8788 MRP.
Introduction . . . .1
Applications and Media Servers . . . .1
IP Multimedia Subsystems and Media Servers . . . .2
Charging . . . .4
Efficient utilization of Media Servers . . . .4
Alcatel’s Media Resource Processing Products . . . .5
Alcatel solutions . . . .5
Conclusion . . . .5
Abbreviations . . . .5
References . . . .6
Acknowledgement . . . .6
Biography . . . .6
1 ALCATEL
Introduction
Traditional service providers are moving away from Time Division Multiplex (TDM)-based circuit switched networks and transitioning towards packet-based networks for scalability, efficiency, and flexibility. Service providers have relied upon TDM technologies with SS7 signaling to provide Intelligent Network (IN) services such as prepaid calling, network announcements, and other services. All these services require some form of media such as announcements, updates to customer accounts, etc. that need to be served/played to the end users. Computer-Telephony Integration (CTI) has introduced a new set of services that can address a wide range of markets/applications. Next-generation services are rapidly evolving supporting voice, video, and data multi-media service-bearing capability with Quality of service (QoS) assurances.
A Media Server (MS) is a network entity that may be shared across several applications and provides common media resources such as Text-To-Speech (TTS), Automatic Speech recognition, Video Conferencing, etc.
A Next Generation Network (NGN) [1] is defined as a packet network in which transport elements such as switches and routers are logically/physically separated from the service/ses-sion control plane supporting multimedia services. The prime characteristics of NGN services are:
a) ubiquitous, real-time, multi-media communications, b) more “personal intelligence” distributed throughout the
network,
c) more simplicity for users and personal service cus-tomization and management.
As a result, the NGN enabling such advanced services will have more distributed intelligence among the network ele-ments. So the NGNs, due to their flexibility and an opportu-nity to rapidly develop new applications/services, are the key enablers for the service providers to attract and/or retain the most lucrative customers for revenue generation.
The NGNs are rapidly converging over packet-based IP works, resulting in more efficiency and flexibility in sharing net-work resources. SIP and H.323 are the emerging and dominant protocols for signaling and call control. The telecommunica-tions industry is rapidly moving towards voice and video over IP technologies. As a result, the Media Servers that incorpo-rate such flexible protocols can also be the core of broadcast for commercial insertion, program playback, and any combi-nation of recent applications. So media server applications are continuously expanding beyond traditional just limited to IN applications such as Prompt and Collect, Interactive, Voice Response (IVR) or simple announcements.
The IP Multimedia Sub-System (IMS) of the 3GPP provides a framework for enhanced and distributed service delivery over IP, independent of access technologies. The IMS acts as a plat-form for delivering any conceivable multimedia service and pro-vides a generic architecture, which contains several common elements that can be shared among several applications. As a result, applications can be efficiently developed and rapidly delivered to the end users by sharing the common elements/fea-tures such as user-profiles, billing, security, etc. IMS-compliant Media Servers will comprise the Media Resource Function Con-trol (MRFC) and Media Resource Function Processor (MRFP) that provide a decoupled and flexible architecture. They can also serve requests from consumer’s IPTV set-top boxes and mobile multimedia terminals/browsers for additional video play-out and streaming. As a result, the Media Servers are poised to play a significant role in triple-play services architecture (TPSA).
This paper gives details of a number of key aspects in the evolution of Media Servers and what they can do in the con-text of IMS/NGN services. Emerging protocols and standards that fuel that evolution are discussed briefly, along with their impact on media server architecture.
Applications and
Media
Servers
Today’s service providers are striving for differentiators in a highly competitive market by providing new services
Annoucement Server Contact Center Calling Card & Prepaid
Enterprise collaboration application IP-Centrex
Audio & Video Conferencing Interactive Voice Response apps Directory services
Location based services
Voice mail and Unified messaging Network hosted Games Intelligent call routing Voice Portal
Voice activated calling Call Parking & Forward
DTMF Detection
and Generation one/PlaybackAnnoucement Audio/VideoBridging Audio/VideoTranscoding RecordPlay/ Automatic speechrecognition Music onHold SpeechText to
reducing operational costs. To help achieve those goals, tra-ditional telecommunications service providers are moving away from Time Division Multiplex (TDM)-based circuit switched networks and transitioning towards packet-based networks for scalability, efficiency, and flexibility. This tran-sition will help enable advanced services that merge voice, video, and data.
With the advent of IN applications, it has become imperative to have dedicated media serving network elements for applica-tions such as announcements or Interactive Voice Response sys-tems. So voice can be considered as the
primary medium for early Media Servers. Figure 1 depicts major current applications in the telecommunications services that utilize some form of media serving capability.
In most of the applications shown in the table, announcements are played as deemed necessary at any point from session initiation to its closure. For example, in a contact center, the end-user may be transferred to another agent in mid-call with greetings or music playing during the transfer. In all such services, a Media Server is required, streaming media in the form of audio or video depending on the receiving terminal capabilities. Convert-ing text to audio on-the-fly for playback to callers is known as Text-To-Speech (TTS). It is also a key feature of current Media Servers.
Web technologies based on HTTP and XML have enabled the delivery of hypermedia. XML-based extensions
such as VoiceXML (vxml) [2] and CCXML [3] have become industry standards for specifying telephony control, confer-ence control, and voice user interfaces. Fundamentally, a voice browser fetches the required VXML page for interpre-tation from a Web Application Server (WAS) as depicted in Figure 3 and can prompt the Media Server to stream audio to the end user.
Another important feature that a Media Server can provide is bridging. With bridging, several audio and video sessions can be mixed appropriately such that video/audio conferencing is enabled for the end-users.
Triple-play services have now become a common offering from several service providers. As a result, streaming video clips for either self-service (or auto-mated call-attendant) branding or interactive TV have become a reality. Figure 3 depicts one such application, where the end users can interact with TV; this serves as an illustration of “beyond simple audio streaming” appli-cations. In other words, the scope of Media Servers is not limited to traditional applications such as Prompt and col-l e c t , I n t e r a c t i v e Vo i c e R e s p o n s e ( I V R ) , o r s i m p col-l e announcements.
Media Servers play a crucial role in developing such NGN services. Figure 4 depicts an NGN with generic Application Servers and their corresponding Media Servers. The first-generation IP Media Servers are directly incorporated into the Application Servers (AS), leading to a tightly coupled Media Server with the AS. This approach is effective for port-ing legacy applications to IP environments quickly.
However, the tight coupling of MSs with ASs leads to reduced flexibility in sharing media resources across the service control plane. As a result, next generation architectures have predomi-nantly separated the IP Media Server from the application. The initial trend has been to develop HTTP-based markup language control, such as Call Control XML (CCXML), and Voice-XML. This decoupling requires some sig-naling for interaction between ASs and Media Servers and is discussed later.
IP Multimedia Subsystems
and Media Servers
The IMS architecture of 3GPP [4] provides a robust and flexible way of providing services over an IP (packet) network. The key value of the IMS
Telephone Carrier HTTP Corporation Voice XML Interpreter Consumer or Corporate Web site
Figure 2: Voice browser for applications
Media
“Thanks for calling. Can I answer any questions
about the mobile phone you’re interested in?”
Consumer answers
555-1212
1st leg of the audio call Click on remote: “Call our operators now” Setop box (RTP mixer) Enter your phone number 555-1212 Includes information on page viewed Network Call routing and Find available agent Ring agent desktop, with CTI Figure 3: Application of Media Servers in video
3 ALCATEL architecture stems from the fact that different
applications/services can share a common set of elements such as user-profiles, authentication/authorization, billing, and media resources across elements, as shown in Figure 5. This is in contrast to the current (legacy) architectures, where the applications are built in a monolithic fashion and the above-mentioned common elements are implemented separately for each application, leading to inefficient resource utilization.
As per 3GPP IMS Release 6, Figure 6 below shows the inter-faces of MRF, where the MRF can be split into a Media Resource Function Controller (MRFC) and a Media Resource Function Processor (MRFP). The MRFC supports services such as con-ferencing, announcements, or bearer transcoding. Ideally, as per IMS, the MRFC will process the SIP messages received via S-CSCF and uses Media gateway Control Protocol (H.248 MEGACO) instructions to control the MRFP. However, efforts are under way to promote a SIP/XML-based protocol between
the MRFC and the MRFP. In addition, MRFC is also responsi-ble for sending accounting information to charging/billing ele-ments.
The MRFP provides resources as instructed by the MRFC, and its major functions are
a) serving media streams for announcements, etc., b) mixing of incoming media streams,
c) media stream processing such as transcoding.
In all cases of Application Server control, all session con-trol requests passed between the Application Server and the MRFC are sent via the S-CSCF using the ISC interface and the interface of the Mr reference point. MRFC addresses are made known via peer-to-peer arrangements within the IM CN sub-system. IMS Release 6 does not explicitly specify the interface between AS and MRFC; however, most implementations have a SIP (dotted line) interface between MRFC and AS.
Even though H.248 has been recommended as the proto-col at the Mp interface between MRFC and MRFP, this has not yet been widely adopted by the industry. SIP protocol exten-sions that give the applications fine-grained control of the media and call control services are currently under development. These help establish connections and inter-face specifications implemented by XML. But unlike VoiceXML and CCXML, there is as yet no widespread industry acceptance of a standard/uni-versal media control protocol. SIP has gained ground in the industry and has been recently catching up with its predecessor, H.323 and its variants.
Basic network media services can be provided through the Netann scheme as outlined in : basic announcements, IVR, and limited bridging [5]. For more advanced fea-tures and mid-call trigger control, Packet Network Application Server Application Server “Native” VoIP Application IP Media Server PSTN
Gateway ControllerBorder
PSTN VoIP
IP Media Server IP Media Server
Voice XML
& CCXML ControlProtocol Figure 4: Transition towards flexible/decoupled IP Media
Servers from Application Servers
IP Network PSTN/PLMN AS AS MGCF CSCF HSS MRF Application Layer SIP SIP SIP SIP MGW Diameter H.248 Session Control Layer Connectivity Layer
Figure 5: Layered IMS architecture
AS S-CSCF MRFC SIP Mr Mp ISC MRFP
Figure 6: Simplified view of the MRF in the IMS network and the relevant interfaces
work is under progress on XML/SIP-based protocols to provide a mechanism to control sessions in a Media Server and the asso-ciated objects (e.g., audio clips). Such a control description can be carried over SIP INFO messages similar to VXML but with additional features. For Video, Real Time streaming Protocol (RTSP) has been the de-facto IP-based stream/media controller. For further information, the reader is referred to the SIP work-ing groups of the IETF.
Charging
The IMS architecture supports both offline and online charg-ing systems. Offline chargcharg-ing invokes the collection of call data records (CDRs) at the Charging Collector Function (CCF) after the session is torn down. On the other hand, online charging stipulates that the Application Server contact the online charging function and, based on the policies established, the session/call is allowed to proceed to access the resources.
Figure 7 below depicts the interaction of various elements with the CCF. As a result, the MRF interacts with the CCF for any charging events that may have to be passed on to the CCF. As the IMS session traverses and utilizes resources such as media resources, the pertinent elements will generate appro-priate Accounting Requests (ACRs) using Diameter. These ACRs will be appropriately processed by the CCF, thereby gen-erating Call Data Records (CDRs) for the billing system.
It should be noted that ACRs can be of two types depend-ing on whether they are IMS-session related or IMS-event related. The former relates to the session depending on the point of ACR generation during the session (start, during, stop). Non-session related ACRs are called events, and they trigger the CCF to generate corresponding CDRs. On the other hand, session ACRs cause the CCF to update corresponding CDRs. The MRFC sends session ACRs.
Online charging, on the other hand requires a different approach. Figure 8 shows the IMS online charging elements. Note that only the elements that are relevant to this paper are depicted. When a UE requests something (resources), the AS/MRFC contacts the Event Charging Function (ECF) through the Ro interface before delivering
the service to the user
.
The ECF in turn can select either a) immediate event charging or b) event charging with unit reservation, depending on the type of resource and policies in place. In the former case, the ECF deducts the appropriate credit from the user’s account and grants the requesting ACRs from MRF/AS. In the latter case, the ECF sends back the allowed time or data volume that can be served to the requesting end user and updates the credit available to the user. This model is suited to a dynamic session where the amount of resources cannot be pre-determined. An IMS-complaint MS should be able to execute the above function-alities.
Efficient utilization of Media Servers
SIP provides the versatility to allow distributed application server components to deliver advanced intelligent services to remote clients. As a result, several key benefits such as increased scalability, fault tolerance, and load balancing capa-bilities can be obtained. However, control of distributed media resources remains an open issue. Most often media resource control is dictated by the inherent characteristics of a network deployment and/or the specific media needs of an application. To gain efficiency and enhance utilization of media resources, dynamic policies have to be deployed. These poli-cies can be determined by:
a) network load conditions,
b) user subscription profiles including QoS, and c) service provider policies.
A fundamental requirement for increasing efficiency and sharing media resources is that both the AS and the MRFC keep track of the usage of media resources, including capacity and capabilities. Current Media Servers lack support of such capabilities, even though AS and CSCF entities do implement some form of Service Capability Interaction Management (SCIM).
The ability to trigger services is a key feature of an IMS net-work. There are three potential trigger handling areas: user preferences, network policy/preference, and service orchestra-tion. Of these, user preferences are well defined in IMS as the
Billing System CDRs RF ACRs CCF MRFC AS S-CSCF
Figure 7: Billing and off-line charging interaction in the IMS
Billing System CDRs RF ACRs CCF MRFC AS S-CSCF
5 ALCATEL
Conclusion
The convergence towards a packet-based IMS architec-ture enables service providers to share and efficiently uti-lize media resources. Common network elements handling various applications can leverage on the shareability and scalability properties of IMS architecture. Various functions and features that are needed for achieving such flexibility and IMS-compliance for Media Servers are described in this paper. The IMS-complaint Media Servers can be efficiently shared across the applications and therefore help increase revenue across NGNs, mainly stemming from the sharing of common elements and applications.
With a full range of Media Servers, including the 8688 MRF and the 8788 MRP, Alcatel covers the needs of NGN and IMS networks for fixed and mobile operators. It also offers a com-prehensive migration strategy from today’s circuit-switched to tomorrow’s packet-switched networks and complements yesterday’s voice processing by video to provide state-of-the-art voice and video processing.
Glossary of terms and abbreviations:
SIPSession Initiation Protocol
RTPReal-time Transport Protocol
VXMLVoice Extended Meta Language
MRFMedia Resource Function
MRFCMedia Resource Function Controller
MRFPMedia Resource Function Processor
CSCFCall Session Control Function
CDRCall Data Records
MRPMedia Resource Platform
TDMTime Division Multiplexing
ASApplication Server
INIntelligent Networks
CCXMLCall Control XML
MSMedia Server
SCIMService Control Interaction Management ability to base a service selection on preferences indicated in
the Initial Filter Criteria (IFC). For example, a SIP INVITE may be directed to an application server (AS) that provides class features for that user.
Network policies and service capabilities have not been well defined in IMS, and both have been referred to loosely as SCIM. Network policies allow the operator to express a routing preference to a server based on events or states in the network (e.g., server outage or the user’s location in the network). This is different from service orchestration, which is a component specialized in invoking services and handling the events and data flow between them. Incorpo-ration of SCIM in the AS and CSCF will allow efficient uti-lization of media resources.
Alcatel’s Media Resource Processing Products
Alcatel solutions
Alcatel’s 8686 Specialized Resource Point (SRP), dedicated to TDM environments and extensively used in intelligent net-works, has evolved into Alcatel 8688 Media Resource Func-tion, which is positioned as a Media Server for NGN and IMS services. It has a large number of flexible features that make it a versatile Media Server providing voice and video pro-cessing. It supports call acceptance and processing via stan-dard IP signaling (SIP) and uses RTP media processing. Media services include voice prompts, menus, and data (DTMF or speech) collection (prompt & collect), half-duplex speech recording, and the ability to bridge or un-bridge calls during transfer. It is integrated in numerous Alcatel services.
The Alcatel 8788 Media Resource Platform is particularly suitable for the applications involving contact centers and answering corporate market needs. The most important func-tions of the MRP are those that permit operation of the plat-form in a multi-tenant environment:
a) policy management for port allocation and resource alloca-tion on a per tenant basis,
b) tenant-based real-time and historical reporting.
This capability is a key benefit for service providers, allow-ing them to easily partition a customer’s application and act as a hosted application service that generates multiple revenue streams. Enhanced call-routing that involves call qualification, call queuing, and call transfer provides the MRP 8788 (that sub-sumes the MRF 8688) with the capability to offer service providers for the corporate market a unique opportunity to cre-ate several applications that are variations of contact centers and a powerful mechanism for revenue generation
References
[1] http://www.itu.int/ITU-T/studygroups/index.html, Study group 13. [2] http://www.voicexml.org [3] http://www.xml.org [4] IP Multimedia Subsystems, 3GPP TS 22.228, 3GPP [5] E. Burger et. al., Basic Network Media Services,draft-burger-sipping-netann-11.txt, work in Progress, IETF.
Acknowledgement
The authors are thankful to Robert Hemmerich for his valuable contribution.
Girish Chiruvolu is a product line manager for Alcatel, North America (NA) for the MRP 8788 Media Servers
in Voice over IP solutions. In the recent past, he was a research scientist at Alcatel CTO, NA and has worked on several projects involving IP, Ethernet, traffic modeling, QoS, and Multimedia applications. He holds a Ph.D. in Computer Science.
A product line manager for IMS-based products and solutions for Alcatel FSD, NA, John Kaippallimalil has extensive expertise in software engineering, network architecture, and standardization of next-generation communications networks.
Jacek Jedruszek has been working for Alcatel since 1984. After participating in the development of a variety
of software projects, he joined product management in 2000 with responsibility for the Media Resource Function and Specialized Resource Point (SRP).
Alcatel and the Alcatel logo are registered trademarks of Alcatel. All other trademarks are the property of their respective owners. Alcatel assumes no responsibility for the accuracy of the information presented, which is subject to change without notice. © 12 2005 Alcatel. All rights reserved. 3GQ 10001 0030 TQZZA Ed.01
