CORDIS Archive

UMPTIDUMPTI - Anytime, Anywhere, Anybody - UMTS for ALL

Alistair Munro
Centre for Communications Research, University of Bristol
Rm 4.12 Merchant Venturers Building, Woodland Road, Bristol BS8 1UB
United Kingdom
A.Munro@bristol.ac.uk

Summary
Introduction
UMTS - An Enabler for Flexibility
Functionalities for Users with Special Requirements
Enabling Technologies and Trends
Lessons Learned in Trials
Outstanding Issues
Conclusions
References and Bibliography

Summary

It is anticipated that a mobile multimedia terminal will be the preferred personal communications device of the future. The imminent deployment of the Universal Mobile Telecommunications Service (UMTS) represents the first step in achieving this vision. It is an ideal opportunity to realise the complementary requirement: mobile terminals must be usable by everybody. The concurrent emergence of the Internet as a viable framework for control and management as well as user services provides many exciting new possibilities for integrating state of the art distributed processing environments based on the Internet, leading to the ability to create and deploy new services rapidly and efficiently. The UMPTIDUMPTI project has studied the issues of access and usability for everybody through a focus on users with special requirements. We report on the results of trials of applications designed for this user group and the challenges for integration of the Internet and UMTS.

Introduction

The UMPTIDUMPTI project [1] has developed a vision of mobile multimedia services accessible to, and usable, anywhere anytime, by anybody. The Universal Mobile Telecommunications Service (UMTS) plays a key part in this vision as an enabler for universal participation in a universal service. This paper is an assessment of how well the anybody element is being addressed as UMTS develops. It focuses on the technology aspects of distributed applications and broadband mobile communications and how implementing key functional requirements appropriate to people with special needs can lead to the flexibility necessary for wider participation.

3rd generation mobile telecommunications services are now a global topic, as evidenced by the high level of activity in the 3GPP, in which the ETSI SMG/UMTS projects play a significant part alongside their counterparts in the US (ANSI) and Japan (ARIB). The process being played out in 3GPP is arguably similar to the one that took place in the fixed network, where the economic pressure to integrate and get rid of incompatible regional standards and technologies had to be reconciled with the impossibility of short-term migration to a common global base. It is focussed on the innovations that are necessary to ensure that a mobile terminal has at least the potential to work all the time, everywhere for the most basic services.

The EU member states have committed to provide UMTS services by 2002 and operators have, accordingly, started to build test networks based on first generation UMTS air-interface products. The first services to be deployed will undoubtedly be those that will relieve the, by then, severe overload on existing GSM networks of circuit-switched voice and data. The existing 64 Kbit/s ISDN/SS7 infrastructure can be reused up to a point, retaining all the well-established management and control systems as well as value-added services that are coming into the mainstream with GSM Phase 2+.

Nobody imagines, however, that this is the future for the infrastructure. With a long-term expectation that untethered mobile access will replace the fixed connection; and considering the impact of convergence of computing, broadcast entertainment, information technology and consumer electronics on the demand for bandwidth at the access point; an infrastructure that can create, manage and deliver multiple information streams from 100 bit/s to multiple Mbit/s between the network and the terminal is essential. In the early stages of UMTS standardisation, it was stated that B-ISDN, (i.e. networks and services based on ATM), would be the partner transmission network meeting this requirement. In 1997 the partner network became the Internet, with retention of established SS7/IN control and management procedures. At the time of writing it is by no means clear how the Internet can support the very demanding constraints on latency and synchrony in the radio access network up to the mobile switch and beyond to the inter-network gateways and fixed network interworking functions. A gradual evolution is therefore more likely.

This evolution matched exactly the logical development route taken by the UMPTIDUMPTI project. Indeed, it had become rapidly apparent that many functional requirements could already, or only, be fulfilled by Internet connectivity and one or more of the Internet application protocols complemented by traditional telecommunications services. Equally, there were other requirements that could not be implemented on an Internet platform of the mid-90s, in particular mobility and integration of legacy telephony services and equipment.

Internet uptake is perceived, or at least reported, as being variously "huge", "enormous", "massive". From the perspective of UMPTIDUMPTI, the faster the uptake so much the better, however hysterically it is reported. Only the Internet can support the diversity of services across all the diverse mobile communications platforms that disabled users will expect to access.

Internet connectivity is, however, only part of the story. It is easy to get, almost universal now. What is more important is the distributed applications that use it, how well it supports them via appropriate signalling and management middleware services, and the impact of a high degree of personal mobility upon it. We begin therefore by asking: U? M? T? S?, in the next section. This is oriented very closely to the UMPTIDUMPTI work and the questions that it raised in a general sense. Next we look at the specific functional and service requirements, followed by the technologies that were used to implement them, and the results of trials. We then project how the requirements would be implemented if certain trends continued. Finally, outstanding issues are discussed.

UMTS - An Enabler for Flexibility

If the mobile terminal is to become the preferred device for personal communications then it must be accessible to, and usable by, everybody. In this section we examine, in a general way, what would be expected of a truly universal mobile service, posing the questions:

Universal

? - the services are pervasive and the same everywhere, and can be invoked at anytime. This is a technology and operational problem with a solution well within industry's capability for traditional telecommunications bearers. Maintaining this ubiquity for personalised and value-added services is the real challenge;

Mobile

? - manifesting itself in several dimensions. Mobility of services and the resources to support them is a prerequisite for universality. Untethered wireless access - maintaining a call or access to a service while in motion, indoors and outdoors, in different parts of a network or across networks - is a liberator for everybody but much more so for people with special needs who cannot function without continuous assistance;

Telecommunications service

? - what does this mean when the transport network is the Internet with mobility and broadband capability?

Universal?

Many people, one in ten or even more, have difficulty through disability in using any kind of communications service. They may not be able to press buttons fast enough or accurately; or to perceive or understand sound or picture information; or even to physically get to the terminal. Almost everybody suffers occasional problems in these areas but, for the disabled, they are fundamental barriers. For flexible working to be an opportunity for all, solutions to these basic difficulties are pre-requisites.

Thus realising a vision of future services for anybody is focused on:

• Enabling users with disabilities to use mobile telecommunications services - widening participation through improving the opportunity to participate;
• Using mobile telecommunications systems for the benefit of users with disabilities - leading to innovative new services.

Expanding on the first point, disability is difficult to define. People may have permanent impairments of sight, hearing, mobility or cognitive capabilities, or they may be temporarily disabled - a noisy railway station, or someone trapped by a fire in a smoke-filled room. They may have multiple disabilities, which may increase with age - or, indeed, decrease for a growing child. This dynamic aspect, operating in the short term as well as the long term, impacting the able-bodied as well as the disabled, must elicit a timely response from the application and supporting services if we claim to be able benefit everybody.

For the second point, new services must be accessible to and usable by anybody, as they would otherwise perpetuate the exclusion of certain user groups. Different abilities will lead to different solutions, often exploiting the specialised communications skills of a specific group and turning the tables in the extreme case - the majority being excluded! In particular, disability is generally accompanied by dependency on carers. Innovations must include and benefit them as well as the people they care for.

Finally, universal means very definitely that services must be available everywhere. The "anywhere, anytime" slogan has encouraged technology solutions that can (may) do this, i.e. a potential - the vendors, operators and service providers have the burden of actually turning it into reality. This has technology implications that are discussed in the next section. It also has operational obligations following moves by regulatory bodies to assure access for all.

Mobile?

The rationale for the UMPTIDUMPTI project was based on taking solutions already proposed for disabled users and seeking answers to the questions raised by adding mobility as a dimension to the communications environment in which these solutions operated. It was recognised at the outset that there would be four main layers of the communications hierarchy that would be influenced fundamentally by mobility: application, service, (and the middleware providing the environment binding the two together), network and physical. This categorisation is quite vague in practice, especially as far as transparency is concerned. Transparency to location, access method, or component failure is very important for distributed IT applications. However it is difficult to maintain when users and other objects move around and their connectivity varies. It may even be an obstacle to the operation of an application when events related to mobility are concealed. In consequence the boundaries between layers become blurred, with a complex flow of management information between the entities implementing the layers.

Application and Service: the Virtual Home Environment

Universal, in the sense discussed so far, can be interpreted superficially as transparency to the several dimensions of mobility. However, if universal implies that services must be available everywhere, certain elements of the applications and services must be mobile to achieve this. Service mobility in this sense is demanded of both fixed and wireless mobile communications systems.

A user can reasonably expect the service accessed and received to be the same in any network, home and visited. In fact this is not the case: emergency numbers are different from country to country; announcements are made in the local language; value-added services, such as traffic news, are not localised but refer back to the home area. The Virtual Home Environment (VHE) is the framework for configuring the state of the terminal and the services accessible to it to reflect its locality, generally at the moment it arrives in a new network. For users with special needs the VHE can be seen as a supporting middleware component that ensures that they are correctly identified and served whatever their locations.

While services exist within an operational network and are terminated at the access points to it, the application includes those physical elements and functions that transform electronic information into media accessible to human users. The services implementing the application may vary. The terminals implementing the transformation can take many different forms. The application is independent of these: fax, for example can be communicated using G3 protocols in the PSTN, G4 in the ISDN, or the hybrid service used in GSM. Fax terminals are well-known in their most common form but origination in electronic form only and termination in storage on a computer hard disk and printing on a standard computer printer are becoming increasingly common.

By this definition, the basic Internet "service" and the V-series modem supporting it in the analogue PSTN is an application. Eventually it should become a service as outlined below. However it is clear that Web browsing, e-mail and similar functions are the true applications in the Internet as it is now. Web-browsing is basically location independent but personalisation, such as bookmarks or local options, is only present on the local terminal. When a mobile user activates a Web browser, it will not usually be personalised to the users' configurations. It will be the role of the VHE to install these bookmarks and configuration details, thus ensuring the service is identical everywhere. The technology to do this is available already [2] but there are still many realisation options: is it on a SIM card? Does the operator store it? Does the user have a globally accessible filestore where it is held?

Other distributed applications have more complex requirements. If we consider a wheelchair-bound user using an application that can locate routes to a desired destination: possibly affecting the duration of "green-man" periods at pedestrian crossings; identifying lifts, doors to be opened or closed, escalators, and similar obstacles en route. This is absolutely location dependent, supported by reference to physical location supplied by one or more network sources or by visual cues relayed (for example) by an on-board camera. The objects - doors, traffic lights, elevators - must be individually addressable in terms that the user can understand as he navigates towards his destination (i.e. they must have meaningful names - possibly temporary and localised - but in any case not in the form normally used by mobile communications networks). This set of capabilities will be shared with the other people in the locality whose physical fitness, location, and intentions may well conflict with each other.

If our mobility-impaired user returns home (i.e. to his residence), the functions are the same yet different. They are very strongly personalised for the owner and those he trusts. They may be constrained for visitors (or family). Moving away from home again, access to the residence and its familiar devices and assistive functions, regardless of where a user is at any one time, is a lifeline back to a set of personal resources. The home also represents a virtual community including the carers (physically distributed in general but usually fairly close), that are invariably part of a disabled person's life. The personal living configuration is likely to be required wherever the user may be.

Distance is a significant influence, leading implicitly to network delays and latency and a consequent need to coordinate simultaneity and the passage of time. For services sourced in the visited network the distance can be expected to be smaller than those originating in the home network. Finer grain information on position will also be needed. Location-based services are currently limited in resolution although improvements are imminent. These rely on exchange of data between terminal and network and their reliability will vary according to how well this can be done. Thus distributed applications of the type envisaged by UMPTIDUMPTI will have to perform adequately on partial knowledge of the terminating party (or parties), never knowing who is participating at any time, where precisely they are, nor how well they are connected.

These requirements exist independently of the local mobile network service providers (they apply everywhere) but they depend on event triggers and information flows that are supplied by, or relayed by, the currently selected local provider. Thus the provider must have access to the flows that it must forward so that it can determine whether they refer to the home environment or the visited environment. This leads in turn to the need to impose security rules at all levels, for routing of inbound and outbound traffic as well as for operations to be performed on the objects the participate in the application.

The discussion above is an exploration of how the VHE could support the needs of mobile mobility-impaired users. It can be extended to include visually handicapped and the blind by creating audible cues, and to other disabled groups. The personalisation may be terminal centred (transforming the presentation of information) or network centred, (when it is impractical to place functions in terminals). The VHE is still evolving: operators know what they need from it - security, service mobility, integration, fraud prevention - functions that only the VHE can provide across operator boundaries. For the user the requirements have a wide scope and degree of complexity that is challenging to implementers and providers but by no means impossible. The communications technology and middleware has existed in the IT and consumer appliance domain for many years.

Network Access

The need for a cable is an enormous barrier to access anytime anywhere. For simple mechanical reasons wireless access anywhere is essential for users who are not physically able to reach or operate a wired device or socket.

Anytime

is more subtle. For user-originated communications this is easy to understand: "I can communicate anytime I want to"; and this applies to anybody. Much of the potential benefit for disabled users relies on all-the-time availability of mobile terminated access that the user may not be aware of, maybe forming the triggers to activate the services and middleware functions of the VHE.

However wireless access leads to uncertainty about a mobile terminal's reachability and location. Due to an adverse local environment a terminal may disappear from time to time. Communication uplink may be possible while the downlink fails, and vice-versa. There is no solution to these discontinuities: macro-diverse cellular systems or redundant broadband spreading can reduce the likelihood of total loss of contact but only at the expense of allocating resources that may never be used or imposing extra processing loads at key network nodes (e.g. WB-CDMA base stations or slots in a TDMA air-interface channel) that cannot be sustained.

The importance of maintaining connectivity is much lower for packet mode services than for circuit-mode services. At present packet data services are implemented using circuit-switched connections: GSM data, (single-slot as well as the multi-slot mode of HSCSD), provides access into the PSTN and ISDN as well as between mobile terminals. The quality of such services depends on maintaining the air link. The ability to do this is highly variable and disruptions can be frequent. Resources are allocated for the duration of the call (which is generally much longer than a typical voice call) but are used at a relatively low duty-cycle.

The introduction of connectionless packet services such as GPRS and better access to store and forward messaging such as SMS and USSD offers better efficiency at the radio resource. It also gives the client application the opportunity to be flexible about reliability, flow-control, and recovery according to its own needs. The Wireless Application Protocol (WAP) illustrates this approach as well as stressing the need for efficient use of limited capacity and independence to delivery protocol.

The WAP provides optimised application protocols to support browsing (WTP, WML) and avoids the need for a full TCP/UDP stack. The use of proxy servers to provide interworking with the Internet proper is implicit and a push mode is available.

Pushing information, unsolicited by the user, requires an "always on" mobile termination for the pushed data. Terminals supporting WAP will have this (SMS, USSD) and do not need an active Internet listener. Equipping mobile terminals with such a listener raises several problems. If the IP address is fixed then the operator is obliged to be an ISP on behalf of all its users. Mobility within the home network can be solved using GPRS (for example). Other operators into whose domain the user roams are also obliged to be ISPs and, as well, to support Mobile IP to maintain the services. Mobile IP has been developed over the past 5 years by the IETF and is mature in many respects but it is not deployed on a scale sufficient to give operators confidence in using it as a service platform. If the IP address is assigned dynamically and transiently, as it is in dial-up access to ISPs using DHCP, then it is impossible to terminate packets until the assignment is made. Dynamic assignment generally implies that the user cannot be identified by a meaningful name, this being bound to a specific address in the current DNS.

The mainstream version of IP, version 4, uses a network layer addressing scheme that is considered to have a limited lifetime. The "new generation" of IP, version 6, extends the size and scope of the IP address and, possibly more significantly, its semantics. The new scheme includes the telephony number space, allowing telephone numbers to be used as IP addresses or terminal equipment identifiers.

The IETF and Internet users are well aware of the current deficiencies. There are several proposals that take advantage of the extensions of IP V6 to allow hierarchical management (similar to GSM MAP), separate address spaces at network level and global level, and other alignments with cellular network mobility management schemes. Such separations can be extended to WLAN, cordless private telephony and other wireless communication systems.

Physical

The compromises on performance due to the physical environment of a radio-frequency last hop are extensively documented elsewhere and the effects on network access were noted above. There are two main influences:

• Variable error rates of several orders of magnitude, affecting throughput or perceived delay, and integrity; and
• Transient connectivity, as a mobile terminal is handed off progressively from cell to cell, area to area, or operator to operator.

It does not matter whether the network that the physical layer supports is wide-area, (e.g. GSM, or TETRA), near-area (e.g. DECT), or local area (e.g. WLAN, HIPERLAN, or BRAN): these influences will always be present. Multi-mode multi-band terminals that are capable of reconfiguring their RF front-ends to adapt to a range of air-interfaces are already available, with the complementary possibility of choosing the best service available at any instant. This situation is already envisaged in the capabilities for cross-infrastructure mobility embodied in IN CS2 specifications, and has implications for the VHE.

The standardisation effort surrounding UMTS and 3GPP has crystallised several targets, including UTRAN (the terrestrial radio access network focussed very much on the WD-CDMA air-interface, although admitting a TDM mode) as well as GRAN (including UTRAN but Generic). WB-CDMA is the enabler of solutions to capacity problems built into narrowband TDMA air interfaces (i.e. GSM). For more detailed technical reasons, a WB-CDMA terminal can maintain several links to different base stations during a session, reducing the likelihood of abnormal call termination in handoff. It is a major benefit if connectivity is maintained at a higher level than possible so far but the relatively low availability of the radio access medium is not easily improvable.

UMTS air-interfaces at the scale of wide-area cellular networks cannot be a one-stop solution when there are WLAN pico- and nano-cellular systems entering the market offering bandwidth several orders of magnitude higher. These will operate privately, indoors and outdoors, probably only in connectionless packet mode. Obviously a mobile user will move within the range of either or both during a communications session and would not expect to have to disconnect from one and reconnect to the other himself. IN CS2 provides mechanisms for invoking the necessary inter-system handover in the telephony domain; other mechanisms are obtainable in the IP domain. This deployment of solutions and products is at a very early stage but there is clearly a path for future development and provision.

Telecommunications Service?

To establish wireless mobile terminals as the preferred personal communicator, UMTS has to offer at least the portfolio of GSM services current when it is introduced and the prospect of near-term availability of services accessible at the fixed network residential gateway.

Technological innovation and billion-dollar investments in infrastructure are well in advance of consumer demand for the wide range of services that are now possible. Advances in high bit-rate digital connections to subscribers (ISDN, variants of DSL) have reduced technology obstacles in the local loop. In a similar way the radio-frequency technologies of UMTS will provide solutions for the wireless mobile domain. The definition of a residential gateway is hot topic just at the time of writing, even if the multiplicity of technologies capable of bringing information to the doorstep is still very diverse.

While the means exists to deliver multi-Mbps to fixed and mobile terminals, a common approach on framing and signalling for control and management has been slow to emerge. The telecommunications tradition offers a wide range of solutions, (Q.931 for ISDN call control, H.245, IN CS2). At present these are seen as the starting point, requiring enhancement to support the anticipated "on-demand" nature of future service use. In this dynamic environment the quality of service (QoS) will no longer be pre-determined by the service provided by the network operator but presented as a requirement from the user or the application.

It is generally accepted that the Internet and its protocols (referred to henceforth as IP) will be the unifier to support the integrated approach whereby in-session requests for changes to the parameters associated with an executing application and service can be propagated between the terminations of the session and to the underlying service provider. Advanced reactive and dynamic optimisation algorithms and protocols can then be executed to acquire and modify the underlying resources to meet dynamic demands.

Such methods are more likely to fail in a wireless mobile network than in a fixed network. The underlying connectivity will change in any case as the user moves, so resources must be allocated and reallocated continuously with a higher expectation of failure. Acceptable quality during the session will be variable, e.g. distortions to voice quality, discontinuous operation, or variable network delay affecting echo-cancellation. In addition users (or applications) may opt independently to change the characteristics of their session in ways that cannot be accommodated by one or more of the other collections of resources participating in that session.

In parallel, there is a move to replace synchronous end-to-end circuit-mode transmission with connectionless packet mode communication in all subnetworks. Clearly this is essential for the IP everywhere approach, in particular for transporting traffic channels. For wireless mobile communications in particular (and as described above), where resources are limited, some of the ambitions for UMTS cannot easily be met if resources are tied up continuously to support conversations that exhibit a significant degree of burstiness. Thus, a packet mode approach that took account of burstiness could offer an opportunity to exploit any underlying statistical multiplexing gain.

In fact the migration to IP for control and management is gaining momentum rapidly. Concurrently, standards groups such as ETSI TIPHON are considering the use of IP for voice and other traditional telephony services. UMTS networks will be in a unique position to take up innovations in IP telephony because they will be unencumbered by a PSTN legacy, although the interworking needs must not be ignored.

The preceding discussion, although very general, is highly relevant to the special requirements of disabled users:

Establishing IP as the basis for control and management as well as end-to-end services matches exactly the support framework identified by UMPTIDUMPTI;

QoS parameter values are likely to be different from those of the majority of users, and will change more often, with the risk that resources cannot be allocated;

Applications for special needs will generally be more complex in their demands for specific or specialised configurations, for multiple concurrent services or for multiparty participation.

Functionalities for Users with Special Requirements

While there is a role for terminals that are specifically adapted to certain disabilities, this kind of solution tends to be exclusive, not inclusive. It also only addresses the minimal set of functions that exist outside the network, when, in fact, a complete solution, (e.g. the mobility application mentioned above), must consider all functional aspects, network centric as much as terminal centric.

The functions were categorised as:

• Local and remote adaptations. Local adaptations include modifications to terminal input functions, such as larger (or smaller) buttons and other widget variations, which tend to be most useful for motor-impaired users. Changes to terminal output functions were limited to adaptations of alerts and extensions of the times to respond to them. Remote adaptations include modification of responses or extension of timeouts when a response from the local terminal or user is expected - these are often too short even for able-bodied users;
• Supporting middleware, including security and directory services, personalisation and localisation functions, message queuing, proxy servers, and synchronisation or transactional functions. Many of these functions must be enhanced to cope with the mobile user to maintain transparency to location, access method, failures, etc.;
• Bearer services, and components that are typically associated closely with conventional synchronous bearers, such as codecs. This category becomes less important if IP end-to-end transmission replaces I, G and V series framing and coding in the long run;

Application communications functions were further categorised as follows, (these are roughly equivalent to the ITU-T I.212 teleservice groupings without the distributive mode):

• Messaging, such as multimedia e-mail, store and forward text telephony and remote access to objects in the personal living space, particularly the home;
• Information retrieval, including access to information repositories such as library catalogues or Web pages;
• Conversational/conferencing, including interpreter relay services for the deaf including real-time text telephony, assistive services for conversation building, or remote monitoring.

The applications, embodying one or more of these adaptation or communication functions, that were developed for trialling included:

• Interpreter relay services for the deaf, requiring real-time videotelephony and text telephony. This is a three-way (at least) conversational service requiring video between the interpreter and the deaf person at least and between all three ideally, audio between the interpreter and the hearing/speaking person, and, if available, real-time text telephony between all three. Improving technology for speech-to-text and text-to-speech and text-to-sign can assist the interpretation process, possibly eliminating the human relay in some cases;
• Multimedia assistive services that allow rapid construction of conversation components by people with speech and motor impairments, and, to a certain extent, those with cognitive problems. Conversation components are retrieved from a multimedia database that may be accessed locally or remotely. There is a conversational flow between two or more parties that requires special synchronisation in the absence of conventional audio/visual cues. The conversation may be delivered using real-time synchronous bearers or it may be done using mail or messaging. Predictive assistants that anticipate the component sequences combined with conversion to/from speech, text and sign can also be used to make the conversation more efficient or to adapt to unimpaired users;
• Library catalogue access for the motor impaired. Although this application is mainly concerned with local adaptations to enable the browser to be used by motor-impaired people, it is built around conventional remote interrogation of the database and incorporates a conversational element using a videophone;
• A SmartHome application allowing a mobility impaired user to control and receive information from devices in the home. A one-way video stream is used to transmit images of selected parts of the home. This configuration is personalised very closely to the users' residence but it is easy to envisage generalising it to the locality in which a mobility-impaired person may be at any particular time, offering the capability to interact in both directions with devices that are entirely outside the telecommunications environment.

Enabling Technologies and Trends

Section 2 introduced UMTS and explained the influence of some of the more general trends and the importance of IP for end-to-end services for the user as well as for the preferred base transport protocol for control and management. Although UMTS is a major enabling technology and exhibits trends as it matures and is deployed, we can project other technological aspects specific to the UMPTIDUMPTI work.

Assuming that IP will indeed provide the uniform network/transport protocol, the bearer service and some application communications functions listed in section 3 can be placed lower in the list of development priorities. Many are already standardised, or will become standards in due course. The trend in this area is to rely on a small number of versatile simple base standards with the necessary configurations and extensions being supplied and matched through open trading or brokering services. This addition of a "level of indirection" is characteristic of all flexible computing solutions, where a descriptor of an entity, (e.g. a service requirement), is presented and resolved to the actual entity.

All the applications were implemented on PCs with IP connectivity (either through GSM or WLAN) and JAVA provided an ideal portable execution environment. The ability to run a client on any JVM/JRE compliant platform was seen as an important long-term benefit, regardless of doubts about performance and the rapid evolution of JAVA, its supporting toolkits and its continuous extension and revision of the basic set of objects.

Using CORBA, or a combination of JAVA client with CORBA server was considered. However, with the likelihood that the server component is just as mobile as the client, the same requirements for execution on a wide range of platforms apply to the server. The IT client/server model used by most CORBA toolkits, and to a certain extent in distributed JAVA applications is poorly adapted to the peer to peer style of telephony applications, requiring contortions in the structure of some implementation components.

The applicability of IN (CS1 and CS2) to support service creation and execution, and the management of distributed call state was also studied although not incorporated in the trial applications. IN CS2 is used already in GSM networks (e.g. to support pay-as-you-go services) and forms the main component of advanced service support through CAMEL. We concluded that the functions of IN would form an extremely valuable complement to IP-based middleware. While implementing IN protocols (TCAP, Core INAP, CS1 and CS2) is technologically trivial, the realisation of the IN functional planes is an enormous task, so our results remain speculative.

Even as UMTS is maturing, enhancement of GSM continues at all levels. Several enhancements are interesting at the application level, although it was not possible to test these within the development timescales:

• Optimising the delivery of a restricted range of services via WAP, as described above;
• A set of core execution environment APIs, termed MEXE, is under development. At present we consider that this is likely to be too low-level to be consistent with the execute-anywhere requirement, although it may well provide the support for a JVM, AWT and JRE;
• The SIM application toolkit, STK (or SAT), allowing development of SIM card resident applications. This will have advantages in simple messaging applications such as the SmartHome, described above.

A common thread in these untested enhancements is that they increase the flexibility of the terminal. We noted earlier that the scope for specialised terminals is very limited. However, a flexible terminal that can be configured to behave in a specialised way is likely to be highly valuable in supporting local adaptations for special requirements, although we have still some reservations about the platform-specific developments that would be required in each case.

Recently announced specifications make it obvious that terminals may not be mobile telephony handsets, network computers, V-series modems and other artefacts in the traditional telephony mould for much longer. Key moves in this area include:

• Bluetooth, presently a wireless RF cable replacement but inevitably evolving into a subnetwork in its own right. It reinforces the trend to separate terminal functions and distribute them to individual physical components. It is an enabler for personal area networking, wearable computing and other innovations;
• HomeRF, in contrast to Bluetooth, a wireless RF pico-cellular network for short range communications. It should be particularly suitable for the home. HomeRF devices are not available yet and the specification is only recently complete. It is similar in some respects to DECT, which has itself taken a long time to appear on the market;
• JINI, a specification for supporting distributed object-oriented processing on embedded devices. In spite of the hyperbole accompanying its announcement, JINI is not new and, at least in public, does not acknowledge the enormous technical, operational and economic obstacles to deploying it. CENELEC TC205 has been developing a specification for home and building electronic systems that is functionally equivalent to JINI with a heritage going back to the late 1970s. Of the millions of nodes in service that use the basis of this specification, problems of cost, performance and interoperability have delayed uptake in the consumer domain.

We hope that these initiatives will succeed where previous attempts have failed.

Lessons Learned in Trials

UMPTIDUMPTI's studies indicated several key areas in which services could be enhanced, essentially confirming the anticipated needs for future development:

• Personalisation - services should behave by your rules - everywhere, and on-demand to adjust to transient ambient problems;
• Multimedia - to combine voice, moving pictures, text or other data streams. The young, the old, the mentally handicapped or non-vocal can all benefit from alternative assistive forms of information and expression;
• Integration of the users' living space in the personal communications space - control and monitoring at a distance, in the home or away from it. It is impossible to overestimate the importance of the physical home environment as an anchor for those who have difficulty functioning away from it;
• Security - to allow participants to be identified accurately, to control what they can use or share and what they cannot. They may be companions or carers - or they may be criminals;
• Design for all - the applications would benefit all users, disabled or not. The accessibility of the applications to all, regardless of impairments, was seen as a key validation criterion.

The trials confirmed also that providing support for special requirements requires a high level of configurability. This applies at several different levels of an executing application:

• Capabilities of the terminals participating in the application;
• Modification of the parameters of the underlying services, which requires configuration operations to be performed on elements in the network as well as the supporting information resources;
• Flexibility in locating functions in the terminal or serving functions within elements of the infrastructure, and modifying this if required as the user roams.

The ease of reconfigurability relies heavily on an object-orientated approach, using JAVA and CORBA models, allowing widget reuse and extension and providing the underlying DPE and middleware. The importance of this method is widely acknowledged in the UMTS community and similar approaches have been used in related ACTS projects, such as DOLMEN, MOVE or MOMENTS.

Outstanding Issues

There is no shortage of future related work in the scope of the IST programme. Some of this can be targeted at relatively simple problems:

• "always on" high bit-rate mobile data services - the mobile termination of IP packets is still unsatisfactory by comparison with the mobile termination of other services. This is likely to be resolved as UMTS terminal equipment matures, although it will remain a problem for the legacy;
• uniform provision and interoperability across multiple communications platforms. This is inherent in IP but implementation of IP across all domains can be complex and costly;
• extending access into domains previously inaccessible to mobile communications services - the personal living space in particular;

Major areas for development are centred on personalisation:

• service mobility and the VHE;
• reconfigurable applications and terminals.

To a great extent the technological basis for these developments exists already, with the main objective being the consolidation of the multiplicity of options. By contrast, there are aspects of IP: Mobile IP and the migration to IP Telephony, that will require significant further research to understand how to integrate them into an environment in which the scale and grade of service required of telecommunications services can be achieved.

Conclusions

The work of the UMPTIDUMPTI project has highlighted the importance of the essential technological and infrastructure topics emerging as UMTS matures. Support for special requirements has been studied extensively over many years and now it is time to take advantage of the technologies and open interfaces to them to build the Information Society for all.

However visionaries in the distributed processing field have sprouted, bloomed, and withered on a rough ten year cycle. Technology is improving all the time and its advances are often mistaken for solutions to fundamental problems in distributed application design, implementation and performance. Foremost among these problems is the liveness and safety of an application involving large numbers of agents and objects, fixed and mobile. Analysis and verification of such systems has a long tradition as a hot topic in theoretical computer science: it is timely that engineers and communications experts turn their attention to the wide range of expertise in this area in realising the vision of UMTS: anywhere, anytime, anybody.

ACKNOWLEDGEMENTS

The author thanks the Commission of the European Union for support through the ACTS UMPTIDUMPTI project.

References and Bibliography

Web site of the UMPTIDUMPTI project: http://www.fen.bris.ac.uk/elec/UMPTIDUMPTI/umptidumpti.html

"User profiles and their Role in the Virtual Home Environment", A. Law, P.Dwyer, T. McNulty, D.Atkinson, Proceedings of the 4^th ACTS Mobile Summit, Sorrento, June 1999

While numerous journal references could be quoted, the pace of change is such that the reader is better advised to monitor the respective Web sites:

European Telecommunications Standards Institute:

http://www.etsi.org, for GSM, TETRA, DECT standards, and information on the TIPHON, UMTS and other projects.

Note: ETSI, GSM, TETRA, DECT and UMTS are registered trademarks of ETSI.

http://www.3gpp.org (the output of 3GPP discussions)

http://www.itu.ch (where ITU recommendations can be bought)

http://www.ietf.org (for Internet developments)

http://www.umts-forum.org (the UMTS Forum in which manufacturers and operators are represented)

http://www.wapforum.org (the WAP Forum).

http://www.omg.org, the Object Management Group

Glossary

Telecommunications terms are frequently referred to by acronym. The table below describes those used in this chapter.

Abbreviation	Translation
3GPP	3rd Generation Partnership Project, a collaboration between ETSI, ANSI, ARIB (q.v.) and other standards organisations in the area of 3rd generation mobile telecommunications.
ANSI	American National Standards Organisation
ARIB	Equivalent of ANSI (q.v.) in Japan
ATM	Asynchronous Transfer Mode, the means of transporting services at different bit-rates in B-ISDN networks.
B-ISDN	Broadband ISDN, an evolution of ISDN (q.v.) to high speed services.
BRAN	Broadband Radio Access Network, (a registered trademark of ETSI).
CAMEL	Custom Applications for Mobile Network Enhanced Logic, a GSM, (a registered trademark of ETSI) specification.
CORBA	Common Object Request Broker Architecture, a specification from the Object Management Group.
CS1, CS2	IN Capability Sets, defining functionality offered by the Intelligent Network.
DECT	Digitally Enhanced Cordless Telecommunications, (a registered trademark of ETSI).
DHCP	Dynamic Host Configuration Protocol, used to verify and connect IP hosts to the Internet.
DPE	Distributed processing environment.
DSL	Digital Subscriber Line, delivering multi-megabit services to the residential gateway.
ETSI	(a regsistered trademark of the ) European Telecommunications Standards Institute
G.series	ITU-T specifications for physical interfacing to digital communications networks.
G3, G4	Variants of the service supporting facsimile in the PSTN and ISDN (q.v.)
GPRS	GSM (a registered trademark of ETSI) Packet Radio Service (a registered trademark of ETSI).
GRAN	Generic Radio Access Network.
GSM	Global System for Mobiles, a registered trademark of ETSI
H.245	The accompanying call-control protocol for videotelephony applications.
HIPERLAN	A WLAN (q.v.) specification from ETSI, (a registered trademark of ETSI).
I.series	ITU-T specifications for ISDN services.
IN	The Intelligent Network, an application using SS7 services to create and execute value added services, route telephone call requests, facilitate charging, and manage resources in digital telephone networks (among many other capabilities).
INAP	Intelligent network Application Part, an IN application protocol using SS7.
ISDN	Integrated Services Digital Network, the first generation of digital telecommunications
ISP	Internet Service Provider.
ITU-T	International Telecommunications Union - Telecommunications.
JAVA	An object-oriented programming language, often used for interactive applications incorporated in Web pages.
JINI	An enhancement of JAVA suitable for small embedded devices.
JRE	The minimum support for JVM to execute JAVA applications.
JVM	JAVA Virtual Machine, executing JAVA programs.
PSTN	Public Switched Telephone Network
Q.931	The call control protocol used in ISDN networks.
SMG	Special Mobile Group, responsible for progressing GSM (q.v.) in ETSI
SMS	Short Message Service, (a registered trademark of ETSI) a store and forward messaging service used in GSM networks (a registered trademark of ETSI).
SS7	Signalling System No. 7, a family of specifications for transporting call control and system management requests between elements of a telecommunications network.
STK, SAP	SIM (q.v.) application toolkit, supporting development of SIM resident applications.
TCAP	Transactional Capability Application Part, an IN application protocol using SS7.
TCP	Transmission Control Protocol, a connection mode, reliable octet-stream service in the Internet.
TDMA	Time Division Multiple Access.
TETRA	Trans-European Trunked Radio Architecture, (a registered trademark of ETSI).
TIPHON	The ETSI, (a registered trademark of ETSI), project concerned with Internet telephony.
UDP	User Datagram Protocol, a connectionless packet service in the Internet.
UMTS	Universal Mobile Telecommunications Service, a registered trademark of ETSI.
USSD	Unstructured Supplementary Service Data, (a registered trademark of ETSI) a data service used in GSM (a registered trademark of ETSI) networks.
UTTRA[N]	UMTS, (a registered trademark of ETSI), Terrestrial Radio Access Network, (a registered trademark of ETSI).
V.series	ITU-T specifications for using the PSTN for data communications services.
WAP	Wireless Application Protocol, specified by the WAP Forum [8], optimising browsing and content delivery in GSM systems.
WB-CDMA	A multiple access mode operating at the physical layer whereby digital information is transmitted in a broadband channel.
WLAN	Wireless Local Area Network.
WML	Wireless Markup Language, part of WAP (q.v.).
WTP	Wireless Transaction Protocol, part of WAP (q.v.).

The next section of this document: ACCORD Solutions for Flexible Working