Meeting the challenge: in the mobile marketplace the customer’s perception of the service is critical. Rodger Bradley examines the role of QoS management as service providers respond to the demands of their customers – Datacoms
NETWORK MANAGEMENT today is very much business focussed. It is about revenue management. With current and future generations of end user devices offering many new services, it is also about maximising returns on investment through effective service management. Control of the quality of service (QoS) is paramount to the success of the new generation of applications and services on offer in mobile communications and the growing “m-Commerce” market.
The “opening up” of telecommunications markets has provided fresh impetus towards the convergence of voice, data and other services, in a highly competitive market. Rising costs and pressure on profit margins demand that service providers, carriers and suppliers are all able to maximise the ROI on infrastructure, equipment and technology.
QoS within connection oriented networks has always existed, but for the almost universal TCP/IP protocol in a connectionless data network world, this has presented some challenges. Today, bringing these worlds together in the mobile communications arena, effective QoS management is fundamental to the successful delivery of voice, video and data services across a network infrastructure.
For the mobile marketplace the customer’s perception of the service is critical to its success, with availability, reliability and response time as principal definitions. For better end-to-end QoS management the service provider must respond to a range of operational and performance challenges, and be able to make rapid decisions based on network demand, congestion, change of service or priority demanded by the customer.
Mobile services have evolved through 2G (voice and SMS messaging), through 2.5G (GPRS), and to 3G/UMTS delivering voice, image and data services from end user to end user, and providing access through a network to multiple application and service types. In all of this, QoS management techniques must ensure cost-effective use of bandwidth and service elements in the same physical infrastructure, delivering multiple services in a reliable, predictable and consistent manner. Using the most effective techniques and having tool-sets at network management centres is essential to the providers’ abilities to make real time decisions, changes in service delivery, or traffic priorities.
DEVELOPING QUALITY OF SERVICE
At the basic level, QoS is dictated by the designed characteristics, capacity and performance of the infrastructure and available bandwidth, whatever services are used for the communications links. In addition, hardware chipsets, microcode, or the communications protocol attributes and policy definitions allow more specific control. The design of the infrastructure provides service quality assurances through resilient links, duplicated hardware and software elements, with alert management and performance reporting tools. All of these feed in to conventional SLA analysis and reporting systems. Maximising the efficient and cost-effective use of the infrastructure to deliver services has to be improved through a “value added” approach, and consider the service in its real end-to-end nature.
Since the end user’ device almost invariably uses IPprotocol, and with packet data services delivered over mobile networks, the technology demands the use of IP QoS techniques as the starting point. The basic or best efforts traffic model provides simple management of incoming packets at each node, where they are queued for onward distribution at each output port, or queue, by IP address FIFO (First In, First Out). At times of congestion, no prioritisation occurs, packets are dropped unacceptable for voice services- and users can obtain more bandwidth by simply ignoring congestion signals. Such an approach cannot differentiate between traffic types and performance controls are only applied to the aggregate flow.
Moving on from the basic model, nodes that use RSVP or DiffServ QoS parameters can regulate more precisely the traffic flow, by applying a weighting to each outbound queue (prioritising). With RSVP, this flow management, reserving resources and signalling priorities from end to end of the route, across each connected node does not however, aggregate traffic types. The third, and perhaps most useful approach is using DiffServ, where traffic flow is not signalled end to end, but on a per packet and per hop basis, and which allows aggregation of traffic types, in addition to weighted queues.
In the conventional world of IP networking, the elements that provide the system of service delivery have some well-defined rules for traffic management, but these are applied, typically, as an overall measure for all traffic types. New applications, and changes to existing services, including greater provision of realtime, voice and video images to more mobile users, is driving changes to QoS management techniques, and existing ISO standards, such as IS08072. Whilst multiple services are already being delivered across conventional networks, this is now happening in the mobile arena using the new generation of cellular communications.
Mobile networks data services use standard circuit switched GSM channels, and delivers the payload at 9.6 kb/s. GPRS data services are theoretically able to deliver at speeds of up to 171.2 kb/s, whilst 3G can deliver up to 2 mb/s. This will enable service providers to deliver new multimedia content, services and applications that can provide a framework for the growth of m-Commerce services.
Value add techniques offered by operational support system (OSS) suppliers, such as policy based networking further improves service performance and customer satisfaction. This approach views QoS from the end user perspective, mapping conventional infrastructure QoS management such as 802.1P/Q, MPLS and DiffServ or RSVP with the ability to differentiate between application services, access rights and billing. Overall, this is a strategy for managing SLAs, rather than simple network management, performance and capacity planning.
The main elements of this approach include: 1. Policy Architecture Models – defining the service, the policies associated with that service, and the conditions and actions associated with implementing the policies.
2. Co-ordination of end to end traffic management practices
3. End-to-End QoS Management for Mobile Services
4. Service from customer perspective – availability and performance
5. Isolation and resolution of service related problems
6. Providing real-time SLA monitoring of SLAs – interdepartmental and external partner networks
7. Benchmarking service performance – by location, by hardware, by network
8. What quality of services are expected from next generation technology
This policy based approach effectively overlays the clearly defined QoS concept and architecture model for 3G UMTS mobile systems. In addition, it overlays the traditional, converged and non-converged voice and data communications networks that have implemented services delivered through TCP/IP protocols.
The capabilities of mobile technology presents a challenge, since the demand for data and other services is outstripping the demand for voice services. The latter uses about 20% of the bandwidth, and provides some 80% of revenue, whilst for data and other services, the reverse is true, where 50 to 80% of the bandwidth used generates around 20% of revenue.
The business needs to improve non-voice revenue, and QoS management improvements will help in achieving the objective. On top of this, voice services are being eroded in the new, highly competitive, deregulated market, providing further drivers for the business to compensate for the decline of voice earnings and improve revenue from all services.
The changes that are taking place in the communications services market is underlined by the expected growth in mobile product offerings across Europe from 18% in 1997 to 38% in 2004, with much of the growth accounted for perhaps, in the transfer terrestrial telephony services.
A little growth is forecast in fixed data, but much of the increase will be taken on through mobile services, further emphasising the need to improve QoS management.
From the business perspective, it is clear that the two basic forms of QoS management are:
Operational: these are in the design, operation, performance and capacity planning based on received, historical data, or SLA targets that are designed in to the service model.
Transactional: based on instantaneous, current data, this provides for service changes and policy decisions that impact customer QoS. Typically affecting application access or selection, with the ability to segregate traffic types, and provide real-time bandwidth allocation. This may lead to a menu driven approach to service delivery, but will demand very sophisticated management tools.
Convergence is a particular focus since data and voice services are derived from opposite sides of the communications platform – voice is driven from roots where the output is known, with guaranteed delivery, having established an end-to-end path. Data traffic on the other hand is much less dependent on the need to establish end-to-end guarantees, since retransmissions, latency and delays have a far less serious impact on the quality of the service.
The arrival of 3G mobile services has given service providers the ability to provide a wide range of new services and applications, with greater depth of content, but the satisfaction of customers on offer will be the determining factor in the success of the next generation of mobile services. Customer expectation is more easily met with flexible, top down, management of the service. This enables greater predictability, and the ability to rapidly, if not instantly respond to changes in demand. At the heart of this is the need (a) to have a strategy that manages the service from end-to-end, and (b) to utilise tools that monitor, control and enforce QoS policies, whilst allowing real time changes to the service, without disruption.
IP protocols are simple, but inherently weak in terms of their flexibility and overall capabilities in performance management, especially in the areas of service quality. In its latest guise, “IPv6”, has taken some steps towards improving QoS management, since the basic TCP/IP packet now contains a 20 bit trafficflow identification. This can be used to specify bandwidth reservation and delay, for particular types of traffic, allowing the packet to be associated with service class, and allow routing to be determined against class of service.
To be effective though, QoS needs to be as vendor and protocol independent as possible, useable by the service subscribers, and very, very granular, where the service, in addition to type, can be managed by packet/frames, or sessions, and classified according to content. This latter could include person to person, VoIP, streaming image, audio, or even by application name. Inspection of the packets being transmitted at this level of granularity should be able to integrate dynamic port, port hopping, application signatures, database name, and so on. A very detailed view, enabling service providers to manage their operating expenses and revenue with much greater accuracy, and bill customers with a greater variety of information, and manage service changes and policies in real time.
The principal benefit for applying better QoS management on mobile services will be to improve the revenue position, against the background of bandwidth allocation by application, service or user access. The hardware platforms’ existing traffic shaping policies, whether at the core or edge device position, are simply not enough to provide effective, reliable, and user acceptable performance management. A range of toolsets is available today that provide this essential “value add” for controlling delivery of voice, video and real time data across networks.
Today, products such as Micromuse’s “Netcool” solutions, or Allot’s “NetEnforcer” and NetReality”, or MetaSolv’s OSS (Operating System Support) tools allow ever more granular control of the service elements.
Such toolsets are widely used by the main service providers, including 02, Vodafone, Orange and T-Mobile, and allow the service provider to monitor not only the infrastructure element alarms, but the service parameters and thresholds built into SLAs. From the service provider’s perspective it is important in the use of any toolset that OSS suppliers have an understanding of the physical infrastructure, in order to implement and customise QoS management tools. In this respect, partnerships with suppliers of QoS management tools have been established, to further improve service management.
Whilst these toolsets allow considerable scope for reactive and pro-active management, they complement the parallel testing of service quality by the service providers, to determine the impact on the end-user of any change in quality, or proposed new service provision. Overall, the view of the service provider is quite clear]y that revenue benefits derive from customer satisfaction and recommendations to other customers. whilst, clearly, customer perceptions and expectations change over’ time.
FUTURE PLANS AND IDEAS
What then of the future, what plans are being formulated, new ideas, tools and techniques generated? Truthfully, the most likely development will crone from changes in customer demand, awareness of what can be achieved and the limitations of current platforms.
The future will undoubtedly be an IP based mobile network offering a wide range of services and applications, from the plethora of end user devices. Current and future generations of hand held devices will provide voice calls as only one of a multitude of services, with devices that range from the disposable to the PDA that is really a mobile workplace. Some devices in use today, such as the Ericsson P800 can already be used by health and public safety services, with rescue workers – perhaps fire services – having access through the device display to follow detailed floor plans of buildings.
In 2001, the “GigaMobile” project effectively defined a way forward in providing a functional architecture that enables the customisation of services based on the end-user preferences for mobility and network connectivity. The overall goal of the GigaMobile project is: “….to develop a proven, yet future-proof, customisation architecture for (mass) customisation of mobile end-user services, and to demonstrate the possibilities and added value of customised mobile services that can be realistically enabled in 2003-2005.”
There is a culture change in progress too, with an increasingly mobile workforce, flexible ways of working, and the success of”m-Commerce”, dependent on the performance, reliability, accessibility and quality of services being delivered. Whatever the future, the tools and techniques to be adopted in managing the QoS, based on industry standards as well as commercial and business improvement processes will be core to such successful developments.
At some point, depending on the sophistication of the end user device, whether that be mobile phone with SIM card, or PDA, it will be possible for service providers to report back to a management centre, the real time performance of the service from the device itself. At the moment, although mobile devices and services are managed with some very sophisticated tools, it will always be the user’s perception of the quality of service that is the defining standard.
COPYRIGHT 2003 DMG World Media Ltd.
COPYRIGHT 2003 Gale Group