Stream service provision to a customer network using a CORBA-enabled residential gateway

Stream service provision to a customer network using a CORBA-enabled residential gateway R.M. Prasad, Y.C. Shou, S. Mohapi and H.E Hanrahan Centre for Telecommunications Access and Services 1 School of Electrical and Information Engineering University of the Witwatersrand, Johannesburg {r.prasad, y.shou, s.mohapi, h.hanrahan} @ee.wits.ac.za Abstract An intelligent residential gateway (RGW) is proposed that defines a standardised means by which Customer Premises Equipment (CPE) in a SOHO network can gain access to Next Generation Network services based on distributed processing. The gateway shifts complex streaming components needed for service provision away from the CPE onto the RGW, thereby keeping the CPE simple while allowing intelligence to reach the consumer domain. QoS, provision and control is examined in the context of the home environment where local traffic should affect service level agreements. Index Terms DPE/NGN, Gateway, QoS, SOHO and Streaming I. INTRODUCTION The deregulation of telecommunication markets and the rapid advances in technology have spurred the evolution of telecommunication networks. Numerous business drivers are encouraging the telco to change to a multi-service, multiparty and mobile provider that is termed the Next Generation Network (NGN). The consumer market for telecommunication is also changing due to increase in broadband access to the home and the growth of the Customer Premises Networks (CPNs). The CPN is growing to contain numerous terminals that use telecommunication networks to gain access to streamed services with QoS requirements. This growth and adaptation of the customer domain is seen as a major influence for providing a standardised means of service provision to SOHO (Small Office/ Home Office) networks. Present service provision to SOHO networks is performed by hiding the complexity of the CPN behind a User Network Interface such as a home gateway. This situation is similar to providing broadband access via a B-ISDN line and providing Quality of Service as far as the B-ISDN modem. Alternative techniques involve incorporating the Customer Premises Equipment (CPE) directly into the NGN architecture, by placing necessary service and connectivity intelligence on these devices. Present techniques fail to address the additional need for internal services in the CPN and to treat this network as an autonomous network where both local and outgoing interactions exist and need to be 1 This work is supported by Telkom SA Limited, Siemens Telecommunication and the THRIP programme of the Department of Trade and Industry. managed. The need to address the CPN as an autonomous managed entity arises from the need to ensure end-to-end QoS for streamed service where part of the connection remains outside the telco control. For service providers to adapt to the growing consumer needs a service deployment framework is required that allows for a standard, open architecture that eases service creation and provision. The use of a distributed processing environment (DPE) that abstracts service and networking layers to reduce complexity for service provision is attractive. The Telecommunications Information Networking Architecture (TINA) proposes a detailed framework for service deployment in an NGN environment. TINA provides business roles and reference points allowing for separation of the service provision from the network infrastructure. TINA s Network Resource Architecture (NRA) is used largely in this paper to house the various functional entities that are required to provide QoS-enabled streamed services in the NGN. TINA s separation of the roles of the service provider from the network connectivity allows for detached focus on communications issues without worrying about issues from higher functional layers such as the Service layer. This abstraction allows the streaming aspects of service provision to be addressed, while the control and management issues are addressed separately [1]. To meet the requirements of streamed multimedia service provision to a SOHO user, in the NGN, critical issues need to be addressed. These include: Stringent Quality of Service (QoS) guarantee of data delivery; The control of multiple stream flows; Modifications of stream QoS requirements; and Point-to-multipoint streams and underlying technology used for service provision [2]. The paper aims at providing a means of integrating the CPN with the QoS assured NGN. We propose an intelligent residential gateway (RGW) that adopts the service and connectivity intelligence that is required in the consumer domain. The intelligent RGW provides a means to control and manage the consumer domain without adding computational overheads to the CPE. The RGW concentrates all the intelligence required for the control of the CPE and by its integration with the DPE provides a means of distributing the intelligence needed to interface

with the provider domain. This paper aims at making the following contributions: Enhances the TINA architecture is enhanced in terms of addressing the CPN, by defining the CPE interactions in the CPN and with the provider domain Provides a means by which QoS control can be provided within the CPN by monitoring and controlling the traffic (on the consumer and provider sides) and using a QoS negotiation mechanism based on QoS agreements between the CPN and the provider Establishes the concepts of user and service hierarchy required to provide prioritisation for QoS negotiation Gives an implementation of the intelligent RGW and how it is to be demonstrated to achieve the above contributions The paper is structured as follows: Section 2 discusses the paradigm of the NGN that is used as a context for the work developed. The use of TINA concepts to define the NGN paradigm is also detailed. Section 3 defines the needs of a NGN enabled streamed service, the role-players and QoS provisioning needs. Section 4 gives a brief review of the TINA NRA and the standard functions of the computational objects required for the purposes of the paper. Section defines the intelligent residential gateway roles and then details an implementation thereof. II. NGN & THE SOHO ENVIRONMENT The fundamental aim involves the incorporation of the SOHO network into the NGN. Integration of the consumer domain into the NGN requires that devices in the CPN can interface with the service and connectivity network intelligence. The management and control of these terminals from a service and communication perspective must be addressed to provide end-to-end QoS from the service provider through a single packet access to multiple users in an autonomous network outside the telco domain. Most of the present NGN paradigms embrace a softswitch architecture involving a switching platform that distributes the functionality of a traditional telephone switch. This paradigm takes the first step in separating the application layer from the network layer. This separation is done by housing all application layer intelligence in the softswitch and catering for the various access methodologies by use of numerous protocols. We assume that the telco has evolved to a multiservice transport network with a QoS guarantee mechanism, and the control and management in the telco domain uses distributed computing. A further assumption is the need for a DPE that spans from the consumer domain to the provider domain. The need for the DPE results from the softswitch (retailer) standardisation of an interface definition language and the further abstraction of the application and networking layers to provide a consistent (protocol independent) approach. A means by which the CPE can communicate directly via the DPE architecture to the network intelligence is needed. CPE communication via a DPE is needed to provide intelligence up to consumer domain. The architecture chosen to detail the interactions of this paradigm is the Telecommunications Information Networking Architecture (TINA). TINA implies a software architecture that offers reusable software components, supports network-wide software interoperability, eases service construction, testing, deployment and operation, and hides from the service designer the heterogeneity of the underlying technologies and the complexity introduced by distribution [3]. The stakeholders that TINA identifies are the initiating user, the service provider (retailer), the network provider and the responding user. TINA implies that software running on the customer equipment allows the user to adequately integrate with the overall service design. Thus TINA incorporates users directly into its architecture and does not try and hide them behind a traditional User Network Interface (UNI). This methodology however requires that additional computational objects be present on the CPE. Furthermore, in a SOHO environment where multiple users compete for access to the service provider and for available bandwidth TINA does not detail any interactions between the devices in the CPN. This paper addresses the integration of a SOHO network as a whole, and not simply as separate user terminals requesting services with the possible future architecture of the NGN. This is achieved by means of a UNI that does not hide the users in the customer domain, but rather provides adequate representation of these users to the provider. The UNI proposed is an intelligent Residential Gateway (RGW) that aggregates the multiple-users in the CPN as a single user to the provider, thereby easing the provider s access and usage functionality. The communication and connectivity issues are represented to the provider as a single user requesting for multiple services simultaneously. In this paper, the SOHO environment is defined only in terms of the streaming communication and connectivity level functional components that are needed as shown in Figure 2. Components in the service layer, which are required to complete the definition of the SOHO network, are addressed in [1]. III. STREAMED SERVICE PROVISIONING A. Streaming requirements in the NGN Increases in the types and numbers of services have begun to appear. The provisioning of these services is complex and is provided at present by utilising connection-oriented networks or allocating large, fixed amounts of bandwidth. The major networks of the future networks are to be IP-

based. The present day Internet scenario however does not adequately address the requirements of future services, due to its best-effort connection nature. Future multimedia applications will require stringent enforcement of quality guarantees and predictable levels of quality and as such certain mechanisms are required: Specification of the required level of QoS using appropriate QoS parameters Translation of specified QoS levels to the actual control and establishment of lower-level network resources to meet the particular QoS level. "!$# % &'(*) +, -.$/ +$# 0$1 4$ 6798 :;=<$6> :? @ O P Q &# '/ T &% 1SR A B C D E E F G D B C H$D I F J D B C FK E L K C D M K N! # % & '(*)2+, -.$/ +$# 03 4 67U8 :;=<$6 > :? @ A QoS provision mechanism links the service architecture with the network architecture and is one link between the abstracted perspectives on service provisioning and network connectivity. Presently TINA s paradigm of QoS provision involves the Consumer and the Provider with a heterogeneous access networks in between, which guarantee QoS provision by establishing Service Level Agreements (SLAs) at the edges of each access network. The above approach makes it impossible to ensure QoS levels with external networks while simultaneously allowing internal CPN traffic to be present. Since the SOHO network is to be autonomous, the external retailer cannot disallow the presence of internally executing services and their ensuing traffic requirements and must provide end-to-end QoS in the presence of traffic in the home environment. This need for end-to-end QoS requires that a means be addressed, by which QoS can be ensured within the home environment. Additionally, a means of implementing streamed services is needed in the context of the NGN. B. Allocation of functions A Stream Management Framework is defined in [4] that provides a means by which the stream interfaces, that give an abstraction to represent the endpoint of a stream, can be identified for control. Providing a mechanism that would manage requests between the CPE and the provider while keeping their interactions separate requires modification of the Stream Management Framework. This management of requests would provide a means by which requests for services from the initiating user go through an authentication, prioritisation and admission control present on the intelligent RGW (discussed later). From a streaming application perspective, Figure 1 shows stream flow abstractions. The diagram gives the overlap of these streaming functional entities with the various domains TINA identifies. The domains are explained later in section. The Intelligent RGW allows for creation, and termination of the stream endflow point, while the TINA service provider (retailer) performs the role of the stream control. Streams are controlled based on requirements specified by the user and the network, i.e. from a service perspective and a network perspective. Stream control is, thus based on factors Figure 1: Streaming components in the context of the project [] such as bandwidth availability, service priority and QoS control. The gateway is required to have intelligence that allows it to identify users (authentication) and their various levels of authority for service interactions and gateway management. The user hierarchy proposed presents a set of profiles that have privileges for service execution and equipment management. For example an administrator, in the user profiles context, could assign priority levels for other users in the local consumer domain. These profiles can deny certain services from being available to some users (admission) and could provide a list of services used and their billing information. A prioritisation mechanism is necessary for the RGW to provide a solution when contention of presentable bandwidth for services arise, which is inevitable due to the multi-user, multi-service attributes of the SOHO network. Prioritisation is to be performed by the RGW by means of a user and service hierarchy that classifies the importance of a connection request and makes choices based on this. IV. TINA NRA COMPONENTS A description of a technology-independent, network control and management system architecture is provided in the TINA Network Resource Architecture (NRA) and Network Resource Information Model specifications. These models define Computational Objects (COs) that house the functions required to provide a service at a communication and connectivity level. The system architecture is described on generic connectivity network level, layer network level, sub-network levels and on a network element level. The TINA NRA deals with accounting, connection, fault and network topology management facets and segments the network architecture into specialised management layers to provide connectivity services. The key management layers are the service management layer (SML), the network management layer (NML) and the element management layer (EML) [6].

3 < 0 4 A Computational objects used in this work and their functional roles are: The Communication Session Manager (CSM) manages a number of communication sessions and can setup, modify or delete a number of stream flows, it is responsible for the conversion between stream flow connections and network flow connections The Terminal Communication Session Manager (TCSM) manages the nodal part of the stream flow and communicates with the CSM to manage the connection setup for the node and know the details of the node bindings that are hidden from the CSM The Connection Coordinator (CC) provides operations for manipulating the associated connectivity session including the connectivity session release operation, which then deletes the CC. V. DEFINING THE INTELLIGENT RESIDENTIAL GATEWAY The proposed intelligent residential gateway is thus defined to be a network device that allows for integration and representation of the users in a SOHO network to the NGN provider. The RGW manages all transactions and interaction within the SOHO network and between the SOHO domain and the provider domain. To allow the integration of the SOHO and the DPE of the access and core networks there must be included in this intelligent gateway the functionalities of the Signalling and Media Gateway. From a media gateway viewpoint, the role performed by the RGW involves standard VoIP voice to packet translation and fax to packet translation (FoIP). All other forms of media translation occur at the end-devices such as PCs or set-top boxes. Signalling functionality involved is greatly increased because of the distribution of intelligence to the edge of the network, and spans from the Service layer to the Network Element layer. This functionality includes the representation of the user domain to the TINA retailer as described earlier, the control of services and their QoS within the SOHO environment, admission control, dynamic bandwidth allocation and control of stream endpoints. Figure 2 gives a representation of the various communication and connectivity level computational objects and domains that are required to represent the various functional components for the purposes of the paper. The traditional Consumer domain as detailed by TINA has been modified to meet the needs discussed earlier. The Consumer domain is split into the CPE domain and the RGW domain. The RGW domain acts as a Retailer to the CPE, which are treated essentially as separate Users. The RGW is able to emulate the association of the users directly with the Provider. The numerous functionalities required by the intelligent RGW are associated with TINA NRA COs and are discussed below. 0 3 4 0 2 / 01 3 < 7: 3 4; :98 8 6 7 3 < = 2 : 6 7: < @ 17? 0 10 > = ; 0 0 0 SSUAP SSM USM SSM TCSM CSMRGW CCRGW Comp TCSM CSM!"# $%& $ # ' $ ( )))) *+, - *.,+,!,* Figure 2: TINA components and domains required Since the aim is to keep the CPE simple, only the functionality of the ssuap with its stream interfaces is left in this domain. For dumb terminals such as phones, a ssuap is created in the RGW domain to represent possible intelligence that the device could provide by means of user interaction. For example, in the case of a call forwarding service the phone itself cannot perform this function, but the ssuap that represents the phone application can be enabled by the user to forward the call to another number. The functionality of the TCSM has been shifted to the RGW domain and acts as a proxy to the various terminals in the CPE domain. This arrangement allows SFEPs to still be associated with the TCSM and terminate at the end terminal. The idea is to create a standardised means by which any device in the SOHO network accesses the RGW. The TCSM component then provides a means of associating the stream flow endpoint in the user applications to the CSM RGW. QoS control and provision is the focus of this paper and this task is placed in the role of the CSM RGW. The CSM RGW is responsible for numerous tasks including: Translation of requested QoS to provided QoS is done through a negotiation mechanism here based on the priority of the service and available bandwidth Receiving the service priorities and other servicenetwork translated issues (e.g. billing information, traffic usage statistics to administrator) from the SSM RGW Monitoring traffic usage in the CPN domain as well as incoming traffic from the external network in order to constantly be aware of available bandwidth Controlling of traffic channels by means of the CC RGW allows for control of Stream endpoints and hence QoS of these services Possible termination of logical service connections are initiated here if a service should terminate in order for services of higher priority to run Managing and interfacing requests from both the consumer and provider regarding changes in a stream flow CC

QoS provisioning requires a way of mapping QoS from a user perspective to a connection level, thereby allowing endto-end quality to be provided by the network for the user. The technique used in the Quality of Service Negotiation in TINA [7] is adopted. The difference between requested QoS and provided QoS must be noted and a negotiation process must be undertaken at the user terminal in order settle on a provide-able QoS. This negotiation involves a means by which the RGW attempts, to the best of its ability, to provide a service with a certain requested QoS. Depending on the priority stamp of the service it is then possible for services being run by other users to be degraded in quality, or terminated entirely if they are on a lower priority level. Alternatively if the requesting user or service is on a lower priority level then the user will be required to lower his required QoS level or deny the service from being run entirely. The prioritisation process is required to be performed at the service layers as information regarding the user profile and the importance or weighting of the service must be established and this information is not available to lower communication layers. The SSM RGW can pass service requests with a priority stamp, on to the CSM RGW that can then perform the negotiation for service bandwidth. This process is not dealt with in the context of this paper and will appear in [1] as it deals with service and usage layer interactions. To represent the Consumer Domain all the communication and connectivity transactions that leave it are aggregated in the CompTCSM that allows for the representation of all the users in the consumer domain. The CompTCSM allows the consumer domain to be represented as a single device to any externally 2 originating service as all references to the ssuap on the responding user side are transparent to the external side. The RGW thus provides a functionality that should remain predominantly transparent to the user in terms of being an intermediate domain between the consumer and the provider. The existence of the RGW functionality only comes into play for normal service execution when there is a contention for bandwidth and certain changes in QoS or terminations need to be made to running services. A. Implementation Present off-the-shelf gateways assume the presence of a Call Agent to provide the call-control functionality. These gateways are usually simple MGCP / Megaco-enabled gateways that are governed entirely by an external Call Manager. sto create an intelligent gateway as required in the context of the project, a standard off-the-shelf gateway is used, however an MGCP-TINA adapter, present on an external PC, is combined to translate the gateway control logic into TINA objects. The TINA objects can then communicate over a CORBA based DPE to service components that lie on distributed TINA-Retailer nodes. The RGW can then conform to the requirements of the NGN by providing distribution of intelligence to the edge of the network and allow seamless integration of the home network with the external networks. VI. CONCLUSION The paper defined a means of providing streamed services to a SOHO network in the paradigm of the NGN with a complete DPE. The use of TINA domains and components in the definition and integration of the SOHO network with a standard TINA retailer was described. The SOHO network was defined in terms of the communication and connectivity functional components required for streaming. An intelligent Residential Gateway was defined that acts as a retailer to the CPE and simultaneously represents the SOHO network as an aggregated object to the TINA retailer. The RGW provides an interface between the CPN domain and the provider domain and manages the interaction that take place between the two domains. A QoS negotiation mechanism involving traffic monitoring, priority-based bandwidth provisioning and transfer was defined. QoS and its provision as an end-to-end issue is important for streamed service in the NGN. The functionality required in the RGW to properly manages and control the services and the QoS agreement were detailed. To implement streamed services, the streaming functionality played by the various TINA components was detailed. This made use of the TINA NRA stream-binding models combined with the CORBA A/V streaming methodology as defined in [4]. A means of implementing the intelligent gateway scenario using an adapted MGCP-enabled gateway was detailed. This technique requires the use of an external PC that houses the intelligence required to allow for the gateway to interact with the TINA objects and hence provide the distributed retailer intelligence. VII. REFERENCES [1] Y.C Shou, et al, Control and management of home networks using a CORBA enabled Residential Gateway, CeTAS group, School of Electrical and Information Engineering, University of the Witwatersrand, Johannesburg, http://www.ee.wits.ac.za/~comms/, June 2001. [2] J. Barr. OSGI: spec basics, interface issues, http://www.osgi.org/news/news12110.html, 2001-- 14. 2 External to the SOHO domain

[3] M. Chapman and S. Montesi, Overall Concepts and Principles of TINA Version 1.0, TINA Baseline Draft TB_MDC.018_1.0_94, TINA Consortium, 199. [4] B. Lai, The Development of a Stream Management framework based on the TINA Stream Binding model, M.Sc. thesis, Dept. Electrical Engineering, University of the Witwatersrand, Johannesburg, South Africa, 1999. [] D. McGrath and M. Chapman, A CORBA framework for Multimedia Stream, in TINA 97 Global Convergence of Telecommunications and Distributed Object Computing, November 1997, pp.239-2243, IEEE, ISBN 0-8186-8337-X. [6] Chelo Abarca et al., Network Resource Architecture Version 3.0, TINA Baseline Draft NRA_v3.0_97 10, TINA Consortium, 1997. [7] J. Rajahalme, et al, Quality of Service Negotiation in TINA, in TINA 97 Global Convergence of Telecommunications and Distributed Object Computing, November 1997, pp.278-286, IEEE, ISBN 0-8186-8337-X. Reji M Prasad received his BSc degree in Electrical Engineering at the University of the Witwatersrand, Johannesburg, South Africa, in 2000. He is currently doing his MSc degree at his alma mater conducting research on the integration of the consumer domain with NGN providers. Yen-chun Shou received his BSc degree in Electrical Engineering at the University of the Witwatersrand in 2000. He is currently working with the main author conducting research on the integration of the consumer domain with NGN providers. His focus remains in the service layers and the interactions between the various role-players in the NGN. Setumo Mohapi is a Research Officer and PhD student with the Centre for Telecommunications Access and Services. His current research interests are in the area of management of load in distributed computing and mathematical analysis. Hu Hanrahan is Professor of Communication Engineering at Wits University. He leads the Centre for Telecommunications Access and Services (CeTAS), a research and advanced teaching centre devoted to improving knowledge and practise in the evolving telecoms access networks and telecoms services.