INTRODUCTION

The concept of next-generation networks (ngns) was introduced to face the reality of the telecommunications industry, characterized by open competi tion among operators due to the deregulation of markets, the convergence between networks and services, and the increasing demand for multimedia applications ¹. This reality creates challenges for operators both in their infrastructure and in their portfolio of services; operators must quickly adopt new technologies and the ability to develop services in a short time, at low cost, to meet market needs ².

Growing mobility needs and customizing services provided to consumers have given rise to more efficient ways to provide services using any tech nology available at any given time; therefore, network infrastructures must provide sufficient resources to offer value-added services. ngns provide a model that allows the operator to improve the provision of resources to integrate all types of telecommunications services under a single network infrastructure adopting internet protocol (ip); this model is referred to as All-ip ². Therefore, operators can move from a model of vertical archi tecture, in which each service they provide has a separate infrastructure (mobile phones, switched telephony, data networks, television networks, etc.) with corresponding types of access, transport, control and application infrastructure, to an independent landscape architecture in which these services are integrated (figure 1).

Figure 1 Horizontal and vertical telecommunication architectures ³  

The telecommunications industry is adopting the IP Multimedia Subsystem (ims) as the reference architecture for the development of all telecommunica tions services for ngns ⁴ that are able to support current telecommunica tions services and that will be useful in the future. One advantage of ims is that it integrates the concept of the convergence of services supported by heterogeneous networks, i.e., networks of different types such as fixed, mobile or the internet. However, ims does not standardize applications but facilitates access to multimedia applications and voice access to different types of terminals and access technologies.

One of the most important aspects concerning the convergence of services is Quality of Service (QoS) because IP networks originally lacked QoS control mechanisms. These networks were designed to provide service delivery without guaranteeing the reliability of the corresponding information, depending on the amount of bandwidth required, especially for services requiring real-time connections (5). Based on recommendation Y.2001 (6) of the International Telecommunication Union (itu), the necessary features were established to ensure end-to-end (e2e) QoS in ngns.

Globally, QoS is currently one of the most researched areas because of its interest to users, operators and regulators ⁷. ims is an IP-based architecture that is constantly evolving and is heterogeneous ⁸ in operator networks; it is characterized by a variety of protocols used in operator network infra structures. Therefore, QoS policy control has become an important research topic ^{9), (10}.

IP MULTIMEDIA SUBSYSTEM

The IP Multimedia Subsystem is a reference architecture for next-generation service provision standardized by the 3^rd Generation Partnership Project (3gpp) and introduced into umts Releases 5 and 6 (March 2003). ims allows telecom operators to offer multimedia services, such as voice, data, video and combinations thereof, under the same infrastructure through packet switching networks based on IP ¹¹. ims is considered a subsystem because it is part of a complete network in which other components are required, such as an access network, to fully function as a service deployment system ¹².

This subsystem is important because it allows different types of access networks to be integrated regardless of technology or internet services combining fixed and mobile networks ¹³; however, for the 3gpp ims described from the point of view of mobile operators (those that support new applications), a body of standardization telecommunication members of the European Standards Institute (etsi) called the Telecommunications and Internet Converged Services and Protocol for Advanced Networking (tispan) adds the necessary ims compatible with networks of fixed-operator (convergence) specifications. The flexibility of this architecture allows for modifications and extensions in the subsystem ¹⁴, and it also simplifies the application design by harmonizing the ability to obtain session control through the Session Initiation Protocol (sip) ¹⁵.

Figure 2 illustrates the ims architecture and its three main levels ¹⁶, which are as follows: the multimedia services layer; the control session layer; and the ip transport layer, in which an ims-based network architecture enables the convergence of different technologies for access networks, such as a fixed network pstn, a broadband xdsl, a wireless lan, or 2G, 3G and 4G cellular mobile networks.

Figure 2 IMS architecture of layered and access networks ¹⁶  

A distinction must be made between "ims core" and "ims" because the ims architecture refers to a "core" or "ims core" (defined by 3gpp) in addition to a non-ims subsystem number (defined by tispan), such as the Network Attachment Subsystem (nass), a Resource Admission Control Subsystem (racs) or the pstn Emulation Subsystem (pes) ¹⁷. The central components of the ims architecture are the Call Session (or State) Control Function (cscf) entities, which are actually sip servers. These entities have specific functions for signaling and routing traffic ¹⁸ and are located at the core level of the Proxy-cscf (p-cscf), the Interrogating-cscf (i-cscf), the Serving-cscf (s-cscf), the Home Subscriber Server (hss) and the Application Server (as).

Figure 3 IMS architecture of the layered components ¹⁹  

Figure 3 shows the different types of components used in the ims architecture ¹⁹; the main elements found in a basic solution are the CSCF entities. Simi larly, various functional entities for managing fixed and mobile networks are defined, and some entities are responsible for contacting the transport network to ensure QoS and prevent misuse of the services provided.

Different protocols are used by these components according to their signal ing function, such as session control through the Session Initiation Protocol (sip), authentication control via Diameter or policy control by the model Open Policy Service (cops); the components have been standardized by the Internet Engineering Task Force (ietf) for the purpose of adopting open protocols used in the tcp/ip architecture to create a flexible platform that can be scalable and compatible with legacy networks ²⁰. The specification of the functional architecture of an ims subsystem can be found in 3gpp 23.228 ²¹, where interfaces, protocols and applications that can be offered to users under the ims network are defined.

QUALITY OF SERVICE

According to ITU Recommendation E.800 ²⁷, Quality of Service (QoS) is defined as The totality of characteristics of a telecommunications service that bear on its ability to satisfy stated and implied needs of the user of the service. ims architecture focuses on ensuring that the applied QoS policies between specific requests for applications (Session Initiation Protocol - sip / Session Description Protocol - sdp) and multimedia flows (Real-time Transport Protocol - rtp) are defined by the network operators according to the type of business and service requirements at the application level that are used to manage the network resources and enhance QoS ²⁸.

In IMS QoS, control plays an important role through a series of indicators that allow for evaluations according to service type, such as bandwidth, e2e delay (delay), delay variation (jitter), data rate and bit error rate ^29). Within the context of an IP network, a service refers to the overall traffic handling of a client through a particular domain; a service is useful only if it meets the requirements of the end user ³⁰. The organizations 3GPP and tispan (Telecoms & Internet Converged Services & Protocols for Advanced Networks) define a set of four classes of QoS for transport networks ^19). Corresponding to the ims services, these requirements also apply to the interconnection of networks that are IP-based (ip-can) access networks and ims. These four classes are listed in table 1.

Table 1 QoS classes for transport networks ¹³  

An e2e multimedia session can traverse a series of heterogeneous admin istrative domains in NGNs, in which the control system of policies must be able to guarantee the QoS resources in all domains involved (Figure 4). Each domain defines its own mechanisms and policies for the provision of QoS depending on the technologies that are accessible to the operator; however, in an e2e session, a mutual negotiation of Service Level Agreements (slas) between the domains involved is required. An sla is a formal contract ³¹ negotiated between two parties that defines the commitment levels associated with network performance and service responsiveness. The two parties can be a user and an operator or two operators, in which case one takes the role of customer to purchase services from another provider ³².

Figure 4 QoS management across domains ¹³  

Policy-based QoS Architectures

To provide QoS in NGNs, the IETF organizations ETSI TISPAN and 3GPP have defined different architectures for the control of data flows. The ietf describes a policy framework through RFC 2753 (39) (Figure 5), which sets policy rules defined as models that become network configurations or devices in an administrative domain. These rules are stored in repositories called Policy Decision Points (pdps) and Policy Decision Functions (pdfs); these repositories recover the rules of appropriate policies in response to requests for policies that are generated by the QoS requirements of services ⁴⁰. The Policy Enforcement Point (pep) refers to the point where a server enforces policy admission control and policy decisions in response to a transaction request from a user who wants to access a resource on a server network.

Figure 5 IETF Policy-based Admission Control architecture ⁴¹  

The ETSI TISPAN architecture developed its own QoS assurance through the standard ETSI ES 282 003 ⁴²; the architecture is called the Resource and Admission Control Subsystem (racs, Figure 6), which is a subsystem respon sible for NGN control element policy, resource reservation and admission control. It is the main component that interacts with the access network and the core network that carries a service that can affect the priorities of the packets through the DiffServ protocol and book resources with RSVP ⁴³.

racs is the logical element of the largest network that allows for interaction between the service layer, the transfer functions of the control resources and QoS support within the respective NGN ⁴⁴.

Figure 6 ETSI TISPAN RACS architecture ⁴¹  

3gpp is defined in ts 23.203 ⁴⁵ and specifies one policy-based QoS called PPC (Policy and Charging Control, figure 7) from the standpoint of the mobility architecture that was introduced in 3GPP R7 (46). The entity PCRF (Policy and Charging Rules Function) is responsible for policy and charg ing control; the Execution Policy and Charging Function (pcef) associates reference points to them ⁴⁷. The concept and the QoS architecture used by the 3gpp specification are described in 3gpp ts 23.107 ⁴⁸. The concept of end-to-end QoS architecture used by the 3gpp specifications is described in 3gpp ts 23.207 ⁴⁹. The pcc architecture is flexible and applicable to a variety of services, access networks and load models; however, it is not well adapted to multimedia service requirements that are dynamically negotiable with QoS according to changing conditions in the network ⁵⁰.

Figure 7 3GPP pcc architecture ⁴¹  

QUALITY OF SERVICE MANAGEMENT

To implement the various policies of quality of service management archi tectures described above, it is necessary to make a study of the parameters involved in ensuring the selected service according to 3gpp ts 23.107 ⁴⁸, 3gpp ts 23.203 and 3gpp ts 23.207 ⁴⁹.

Quality of Service Attributes related to Policy Desition Point

The Release 99 of 3gpp recommendations defines QoS attributes such as Traffic class (Table 1), Delivery order, SDU format information, SDU error ratio, Maximum SDU size, Maximum bit rate for uplink, Maximum bit rate for downlink, Residual bit error ratio, Transfer delay, Traffic-handling priority, Allocation/retention priority, and Guaranteed bit rate for uplink and Guaranteed bit rate for downlink ⁵².

QoS parameters related to policy control

The following mechanisms (figure 8) are managed by the pcrf ⁵⁴ to ensure the traffic over the lte (Long Term Evolution) and lte-Advanced networks ⁵⁵:

Figura 8 Policy and Control QoS Parameters ⁵⁵  

• QoS Class Identifier (qci) is a scalar valor that control packet for warding treatment (for example, scheduling weights, admission thresholds, queue management thresholds, link layer protocol configuration and so on) ⁵².

Table 2 QCI standardized values ⁵⁵  

1. Allocation and Retention Priority (arp) It is used as an indicator of priority setting processes or modification of an EPS bearer service enabling the network to decide whether to accept these processes depending on available resources ⁵⁵.

2. MBR (Maximum Bit Rate) indicates the maximum number of bits delivered to the network or by the network within a period of time ⁵².

3. Guaranteed Bit Rate (gbr) indicates the guaranteed number of bits delivered to the network or by the network within a period of time ⁵².

DISCUSSION

The central problem is to provide consistent end-to-end QoS to an IP ser vice in such a way that the requested QoS requirements are met when the deployment of a service infrastructure involving two or more operators, being autonomous networks whose administrative domains, are managed according to their own policies.

Although operators must agree on QoS requirements for a particular service among a set of IP services, operators do not configure their networks in the same way. On the other hand, operators and service providers have their own internal topologies and QoS mechanisms that depend on their devices and other management requirements that are non-technical, and it is ob served that the ims has become the defacto standard in the deployment of lte networks, due to its open interfaces for deploying converged services. Further studies should consider this reality taking as baseline the Policy and Charging Control framework defined by 3GPP. That is why it is necessary to develop a guideline that allows operators to maintain a consistent level of QoS interconnections without depending on the complexity of the network.

Similarly, the regulatory aspects regarding the quality of service must be taken into consideration by the regulator. In the case of Colombia, the CRC has been working on issues relating to the adoption of next-generation net works by the industry group ⁽⁵⁶⁾, which aims to promote¹ "...cooperation among all industry players involved in the development and deployment of ngn as well as to track the evolution of the same..." in order to guarantee sla compliance for the user and address the implications of the adoption of these technologies in terms of the convergence of services and markets.

CONCLUSIONS

The evolution of operator infrastructure has led to the implementation of softswitches that allow migration to an All-IP network. However, the IMS architecture has been dominating industry due to its open interfaces for the deployment of converged services. The main challenge is to provide a consistent end-to-end Quality of Service through an IP service, such that the requested QoS requirements are satisfied when the deployment of a service involves the infrastructure of two or more operators, which are autonomous networks whose administrative domains are managed according to their own policies. Although operators must agree on the QoS requirements for a particular service among a set of IP services, operators do not configure their network devices in the same way because they have their own internal topology and QoS mechanisms that depend on their network devices and other non-technical management requirements.