chap4

In this chapter, we present our analysis of five case studies: private business radio (PBR), public paging, fixed links at 1.5GHz and 7.5GHz, and mobile satellite systems in the UK. The issues addressed in each case differ considerably, covering the competition between and the technical efficiency of different mobile services and the costs of congestion and migration on fixed link users. In each case, our analysis has been based on discussions with and information obtained from users and user groups, and information obtained from published sources.

This case study examines a number of aspects of private business radio (PBR) in order to understand better how the spectrum for this service might be managed in the future. Three key issues are assessed:

In order to understand how PBR is positioned in the overall mobile communications arena, it is helpful to understand how the current users employ PBR and why they have selected PBR over other alternatives. A market segmentation of PBR showing the number of users by user category and their estimated current spend on PBR is provided in Appendix 4.

Alternative mobile radio services which PBR users might employ are mobile data networks, PAMR networks and cellular networks. The reasons why PBR users might consider these alternatives is that they avoid the following problems associated with PBR:

GSM operators are making some attempt to attract PBR users, with both operators having products specifically targeted at this market. PCN operators, whilst not targeting PBR directly, offer very low call rates in certain local situations, which has resulted in them gaining some PBR users. However, neither are taking significant steps to overcome the disadvantages listed above. Nevertheless, the considerable advertising and marketing power of the cellular operators compared to that of the PBR community, may mean that GSM could make greater inroads into the market than the results of a functional and cost analysis would suggest.

Bodies which are known to have made the transition from PBR to shared services include a number of local authorities, a few utilities, various haulage and delivery companies, some roadside assistance organisations and various emergency services. Railtrack is also currently examining the possibility of moving its voice communications onto cellular.

A key issue in the choice between cellular, PAMR, public data and PBR is the service cost. As the tables in Appendix 4 show, the cost of PBR for many users is in the region of Ł200 per user per year (based on Smith estimates), compared to around Ł300-Ł400 for PAMR (based on NB3 tariffs) and public data network users and around Ł500-Ł800 per year for the average business user on cellular (based on information published by Vodafone). The effect of these pricing differentials on future PBR growth is discussed below.

It is difficult to map accurately trends in PBR user numbers because, until recently, sufficiently detailed statistics have not been available. The RA has published a breakdown of PBR user figures for both 1994 and 1995 and comparing these figures the following points can be observed.

1) From 1994 to 1995, there was an annual growth in total number of PBR users of around 3%.

2) There was a fall in user numbers in almost all categories except on-site single channel systems. The latter grew by 22%, or around 40,000 users, between 1994 and 1995.

3) Excluding on-site users from both the 1994 and 1995 figures, then there was an overall decline of around 5% in the remaining PBR users.

It is not known how accurate these figures are, nor how comparable the data for the two years are, and hence this analysis needs to be treated with some care.

There is some industry evidence that the growth in on-site usage has been stimulated by the RA’s recent move to make spectrum available for short-range business radio (SRBR). This is a relatively inexpensive analogue single channel radio with limited re-tuning possibilities in the case of interference, which is well suited to on-site single channel use. As well as traditional uses such as construction sites, SRBR has recently been finding new applications such as in shops and hotels. Some have attributed its success to its simplicity with resultant low cost (less than Ł100 per terminal).

Manufacturers predict strong growth in the SRBR segment. They also note that, at present, this segment is restricted to business use. They predict significant demand from consumers for leisure applications but note that, at present, the licence conditions state that the radio channels are only for business use. Whilst this does not exclude use by leisure users as part of a business application (eg golf clubs), the manufacturers are concerned that the market for personal use of SRBR is not being realised.

For the majority of PBR users which comprise organisations such as taxi companies and on-site users, PBR is positioned as a substantially less expensive mobile communications system than any of the alternatives. For these users, although the relative ease of use of PBR is a significant factor, it seems likely that cost is by far the most important driver influencing choice of communications service.

For the higher value PBR users, such as utilities, PBR provides an ability to control their own mobile communications to ensure it fully meets their, often demanding, requirements for availability. For these users, PBR is often more expensive than shared mobile communications systems, but the advantages of having control over their system outweigh these additional costs. Table 4.1 summarises the advantages and disadvantages of the different services to various groups of users. Note that large users who require coverage over many small and isolated sites (eg stores) are included in the category of short range users.

User Type	PBR	PAMR/public data	Cellular
Large organisation: eg utilities, governments users, railways	For: Control over system. Coverage tailored to requirements. Against: Can be expensive.	For: Relatively low costs coupled with some control. Against: Coverage and functionality poor (digital PAMR should enhance this)	For: Good functionality including PSTN interconnect and low blocking. Against: System parameters and cost difficult to control.

Small users/wide area roaming: eg bus and coach operators, hauliers	For: Control over system. Against: Typically extremely expensive	For: PMR provision at less cost. Against: Coverage may not meet requirements.	For: Good coverage. Against: Typically more expensive than PAMR

Small users/local area roaming: eg taxi companies, despatch	For: Lowest cost provision. Against: Unlikely to be a core competence	For: Provides similar functionality in outsourced form. Against: Typically more expensive.	For: Low blocking. Against: Lack of group calls and typically most expensive option

Short range users: eg construction sites, stores, large manufacturing plants	For: Simple and cost-effective. Against: Nothing significant.	For: Provides similar functionality in outsourced form. Against: Typically significantly more expensive.	For: Nothing significant. Against: Lack of group calls and typically most expensive option

Based on these advantages and disadvantages and the relative costs of the different services, it is possible to conjecture how the use of shared systems might evolve in the coming years:

Based on these assumptions we make the tentative predictions about the PBR marketplace over the coming decade shown in Table 4.2. These predictions assume that users will seek to minimise cost, except where other over-riding issues are apparent. Details for the different PBR segments are given in Appendix 4.

User type	Typical current system	Trends
Large organisation	PBR	Towards PAMR and cellular
Small users/wide area roaming	PBR and PAMR	Towards PAMR and cellular
Small users/local area roaming	PBR	Remaining with current provision
Short range users	PBR (Short range radio)	Remaining with current provision

Were the migration shown in Table 4.2 to occur over the next ten years, this would reduce the overall number of PBR users (including railways) by around 260,000 (i.e. about a 30% reduction in current levels). On current trends, this would be more than offset by growth in the remaining PBR users, particularly in the short range applications.

The above analysis suggests that, although the total number of PBR users may continue to grow, the growth could mainly arise from short range applications, whilst the number of users requiring regional or national PBR channels could fall. In considering the impact on demand for radio spectrum in the longer term, we offer a number of scenarios could be envisaged:

At this stage, it is too early to predict which scenario is most likely, although we note that typically, if more spectrum is "found" applications rapidly appear which can utilise the spectrum.

This scenario will have some significant implications for spectrum management. Short range operation is significantly more spectrum efficient than standard PBR as substantially better frequency re-use can be achieved in a given area. As a result, only around 3-5 short range channels would be required to accommodate the predicted growth in the short to medium term.

In principle, a gradual reduction in the number of standard (ie not short range) PBR users would result in spectrum becoming free within the PBR bands. In practice, given the current high density with which users are accommodated in these bands, particularly in urban areas, the results of the migration for the first few years may be simply to reduce congestion, rather than to free up spectrum. Hence, in practice, re-farming is likely to be a long-term process requiring careful management.

This represents the status quo. Hence, current RA management policy would remain suitable for this scenario, with a continual need to increase the efficiency of use through measures such as the introduction of narrowband systems and encouraging the migration towards the use of trunked systems.

In many ways this scenario is similar to scenario two in that there would be a continual need to increase efficiency. However, depending on the new applications, there may be a need to partition spectrum and hence migration issues could arise.

Given the clearly different implications of these scenarios for spectrum management, it is important that the RA undertakes further work to establish the likely future trends in PBR use.

In considering the relative positioning of different services, some of which are at least partial substitutes for each other, the impacts on spectrum requirements of migration from one to another are an important concern to the spectrum manager. In order to understand these impacts, it would be helpful to have a measure showing how much spectrum each user required depending on which service they chose to adopt. This requires the same kind of analysis as is needed to answer the question: which service is the most "efficient"?

In this section we try to derive such a measure. In doing so we make a key assumption that a user migrating from one service to another will continue to send the same amount of information on their new service as they did on their old service.

There are a number of possible measures of spectrum use or efficiency which might be considered:

Explanation: A measure of the bandwidth required for a single voice channel. In the case of frequency division systems, this is given directly by the transmitted bandwidth. In the case of time division and code division systems it is necessary to divide the bandwidth by the number of channels per carrier.

Advantages: Simple and uncontentious to calculate in most cases. The only exception to this is code division systems where the number of channels per carrier can be difficult to determine.

Ability to compare spectrum requirements across services: Without consideration of the re-use factor and the number of users per channel, use of this measure to estimate the spectrum required by a user if they migrated to a different service has the potential to be highly misleading.

Explanation: The total number of subscribers or users of a certain piece of spectrum is divided by the amount of spectrum assigned.

Advantages: The data to perform the calculation is typically readily available. Re-use is taken into account by considering subscribers on a national basis.

Disadvantages: The main disadvantages are firstly that the amount of information sent per user is not taken into account (eg PBR users tend to generate more traffic than cellular users on average), secondly that the relative "quality" of the spectrum is not accounted for (eg some PAMR spectrum is subject to significant Continental interference), and thirdly that new services are seen as inefficient in their initial years.

Ability to compare spectrum requirements across services: Although this measure can provide the amount of spectrum each user requires on average, the fact that the relative amount of information sent at present varies between different services means that one is not comparing "like with like".

Explanation: The total amount of data, measured in bits, carried within a given area (which may be the entire country) is divided by the total spectrum assigned.

Advantages: This measure better allows comparisons between different systems since it might be assumed that a user will transmit approximately the same amount of data even if moving to a different system and hence the relative amounts of spectrum required is revealed. This measure also takes into account the throughput systems actually achieve as opposed to what they might achieve in theory.

Disadvantages: It is difficult to calculate as information on the amount of data transmitted is not readily available. It will appear low for emergent services and services where the quality of the spectrum is low.

Ability to compare spectrum requirements across services: This measure allows a reasonable comparison of the efficiency of different services.

Explanation: The bandwidth required per voice channel, as discussed above, is multiplied by the total number of channels that are required over a wide area to provide the equivalent of one voice channel in each sub-area (eg the cluster size for a cellular system). The need (in most cases) for multiple channels is a consequence of the interference that would arise if the same channel were to be used in adjacent sub-areas. This approach results in a measure of the technical efficiency of the system over a wide area.

Advantages: It overcomes the key disadvantages with the bandwidth measures and provides an accurate representation of the number of channels which can be used within a given area.

Disadvantages: It does not take into account how heavily a channel can be loaded. It can be difficult to calculate as there is often little agreement over the cluster size actually achieved. It does not take into account significant trunking gains achieved by multi-channel systems.

Ability to compare spectrum requirements across services: This goes some way to allowing the likely spectrum requirements of a user moving to different services to be compared. However, by not measuring what is actually achieved, rather than what is theoretically possible, significant errors may arise.

Tables showing broad estimates of these measures calculated on a national basis are provided in Appendix 4. The numbers used in these tables are approximate and are only used to illustrate the likely outcomes. When considering the results it is important to note that PAMR spectrum is subject to Continental interference which has caused it to score badly on some measures. Also, PCN is less mature than GSM which tends to make it score relatively badly on the measures of subscribers/MHz and traffic/MHz. A summary of the results is provided in Table 4.3.

* For these measures the smaller the value the greater the efficiency of the service.

Considering all the measures for which it was possible to gather sufficient data we find that the measures give dramatically different orderings of services, and almost all the services considered fare well according to at least one of the measures.

In terms of the ability to predict how much spectrum a user would require moving between different services, only traffic/MHz/day seemed conceptually sound. However, this measure was difficult to estimate and a number of assumptions and approximations were used in its derivation. It shows that GSM achieves the highest efficiency, probably as a result of its trunking gains. PCN currently achieves a lower efficiency, but as the network approaches maturity, it is expected that this efficiency will reach similar levels to GSM. PBR only achieves around a quarter of the efficiency of GSM and two thirds that of PAMR, probably due to a lack of trunking gains, although it should be noted that in order to derive the PBR figure a number of broad estimates on channel occupancy (obtained from RA monitoring data) and re-use assumptions were made (see Appendix 4).

Paging transmits a relatively low amount of information per MHz. However, paging users transmit a relatively low level of traffic per subscriber. Dividing the estimates of traffic/MHz by the number of subscribers/MHz given in Table 4.3 gives average traffic/subscriber. Using this information, a user moving from GSM to paging would be expected to reduce their traffic levels in Mbits by a factor of 4,000. Taking this into account increases the "comparative efficiency" of paging to 440,000Mbits per MHz per day, significantly better than the other services.

Hence, these measures would suggest that PBR users moving to GSM or PCN would only require one quarter of the spectrum they currently use. PBR users moving to paging would only require around 4% of the spectrum they currently use. However, in both cases, their functionality might be compromised and hence the value they receive from the spectrum reduced.

In summary, our analysis suggests that the search for a technical measure which simply and uncontentiously allows different mobile radio services to be compared is likely to be controversial, both because it is difficult to produce a single widely agreed definition of the measure and also because sufficient data is often not available to calculate the result. Furthermore the measures are simply technical measures of efficiency and do not take into account the quality of service experienced by the users. It is therefore inappropriate to draw firm conclusions about the relative spectral efficiency of PBR in comparison to other services, on the basis of these measures alone.

PBR spectrum is currently assigned on an application by application basis by the RA’s headquarters and regional offices. An assessment is made of whether there is a requirement for PBR, including, in certain cases, an assessment of the spectrum efficiency of the proposal.. Once the requirement for PBR channels has been verified, assignments are made taking engineering issues into account and incorporating any local factors such as the terrain. In this section we consider some alternative management techniques and examine their potential benefits.

Each of these techniques is discussed in turn below. We looked at each of these different management techniques, but in no cases were we able to discover any quantitative evidence to show the gains which might be achieved with each technique. Nonetheless, each technique has a number of potential merits and further investigation, beyond the scope of this study, is required before one could make recommendations as to future management techniques.

SMOs are bodies which would manage the spectrum of behalf of the RA. At present there are a number of SMOs active in the area of PBR, such as the Joint Radio Company (JRC) which acts on behalf of the gas and electricity companies. In principle, SMOs have the capability to make more effective use of the radio spectrum. Because they understand the needs of their users better than a centralised spectrum manager would be able to, they should be able to make assignments more appropriately and allow more users to be accommodated by providing varying levels of protection to the different users according to their requirements.

We discussed with SMOs in the UK and the US (where they are known as frequency co-ordination groups) whether there was any evidence of the gains which SMOs could achieve. Although they could show a more rapid response time to users, they had not collected any information on any spectrum efficiency enhancements (eg the additional number of assignments they were able to make compared to the RA or FCC). Further work with these bodies would be required to model any efficiency gains that had been achieved.

The RA maintains databases showing who is licensed to use each part of the radio spectrum. These databases are currently confidential and not available for those outside the RA to examine. Some manufacturers and users have argued that if the database was publicly accessible, those who wanted access to the spectrum could examine the database to find frequencies they might use and then suggest to the RA that they might be assigned these frequencies. Indeed, there is anecdotal evidence that despite this access being denied, some users have in the past attempted to suggest to the RA the frequencies which they should be assigned. If such an approach were to be trialled, perhaps in a limited frequency band and geographic area, this might provide evidence of potential gains.

This approach would seek to maximise the number of users by ensuring that, as far as possible, the transmitter power used and transmitter height are at a minimum and that techniques such as CTCSS were adopted. This will reduce interference and hence allow greater frequency reuse. To some degree, this is simply an enhancement of an approach already adopted by the RA. It is not clear that there are any significant gains to be had in this area.

As technology progresses, techniques emerge which allow radio communications to use less radio spectrum than previously. This is often termed "narrowband technology" as the most obvious form of this progress is when new equipment is able to use less bandwidth than older equipment. Typically, the newer equipment is more costly than the old until significant sales are achieved. The RA could mandate the use of narrowband equipment for all existing or new licences in order to speed its introduction. As such policies will impose costs on existing users and it is not clear whether demand growth is such as to warrant such a change, we do not see any merits in pursuing such a policy until the pattern of demand becomes clearer.

At present the RA does not offer PBR users a guaranteed grade of service. Consideration of administrative pricing or the use of SMOs has raised the possibility that users could be offered differing grades of service at differing costs. In this way, users with a strong need for a channel with a high availability could opt to pay higher licence fees in order to enjoy a commensurate level of service. There is no comprehensive measure of the traffic by time of day and hence availability on different PBR channels. At best, the RA knows the number of mobiles per channel, which is likely to provide a poor measure of availability as users’ traffic levels may differ greatly. Thus, in practice, it is likely that at least initially only two grades of service could be offered - either shared or exclusive use.

To take this idea further, the RA would need to have a better understanding of the value to users of improved availability and whether numbers of mobiles could be used as a reasonable proxy for traffic levels or, for example, whether the measure would need to also distinguish the type of user. The former would require a survey of users’ willingness to pay for different levels of service, whilst the latter might involve collecting traffic data for channels with very different numbers/types of users and then analysing the results.

This case study examines the positioning of paging in the overall communications market which includes data services, voice services and technologies capable of offering both data and voice services (eg GSM). In the future this may also include digital broadcasting which has some data capabilities. We were also asked to examine the efficiency of paging. This is discussed in section 4.1.3 above.

There are currently four national paging operators in the UK - BT, Vodapage, PageOne Communications and Hutchison. In addition there are three regional operators. All the operators use the POCSAG standard which has now been in use for many years. Most operators have four 25kHz national channels. Six licences for the new paging standard, ERMES, have been granted to each of the existing four national operators, London Paging and Pagenet. The operators may need to relocate some PBR users before they can deploy their service.

The paging market in the UK had been relatively stable from the mid 1980s until around 1995, with a total of around 800,000 subscribers. However, the introduction of calling party pays (CPP) tariffs have changed this, with annual growth now around 20% and around 1m subscribers in the UK. Operators expect the core 800,000 subscribers, most of whom are business users, to continue to use paging in the future. These are typically users for whom paging fits well with their needs or preferences, who have used paging for a number of years and are thought unlikely to change regardless of enhancements to cellular phones. The new users are mostly domestic consumers who are attracted by the absence of any on-going charges with CPP. They often couple "free" paging with the use of phone boxes to achieve mobile communications. Average tariffs vary from Ł30/month for business users to nothing for CPP users. Growth in CPP is expected to continue for a number of years, increasing the number of paging subscribers by as much as 20% per year.

Paging penetration in the UK is substantially lower than in the US. Some observers have questioned whether UK paging levels will rise to those in the US. This seems unlikely as paging penetration in the US is increased by the cellular tariffing policy which requires that mobile phone users pay for incoming calls. To overcome having to pay for unwanted incoming calls, many mobile phone users bar incoming calls and use a pager as a method of screening call requests. This tariff structure does not exist in the UK and hence paging penetration amongst mobile phone users can be expected to be lower than in the US. Estimates from UK paging manufacturers suggest that around one third of paging users (ie around 350,000 subscribers) also have cellular phones. With a current UK cellular subscription level of around 6m, this implies only around 6% of cellular users have pagers.

The reason that paging, a simple one-way communications medium, has survived and recently grown even though more advanced cellular systems offering two-way communications have become ubiquitous is down to a number of factors, including:

Some operators provide a limited roaming service to other parts of Europe and also information services such as the distribution of financial information.

Cellular is eroding some of the advantages listed above. Coverage is getting better, tariffs are falling, battery life is being significantly enhanced, size is reducing to a level commensurate with some pagers, and services such as the short message service (SMS) allow incoming "calls" to be screened. This would seem to imply that cellular could be a greater threat to paging in the future.

Another alternative is public data networks. These allow paging messages to be sent and also allow a response to be received. This two-way interactive capability is preferred by some users, but it comes at a cost of lower coverage and a significantly larger and more costly terminal. As a result, public data is unlikely to have a significant effect on the size of the paging market.

A summary of the advantages and disadvantages of paging, public mobile data, cellular phones without a data facility and cellular phones using the SMS facility is provided in Table 4.4.

	Paging	Public data	Voice over cellular	SMS over cellular
Advantages	Coverage. Size. Battery life. Ability to screen calls. Low cost. Group calls.	Two way message capability	Ability to reply directly to message.	Full paging functionality plus return receipt and phone facility when required.
Disadvantages	Lack of return receipt. Need to find a phone if call back required.	Size and coverage plus no voice backup unlike cellular	Difficult to screen calls. Coverage not as good as paging. Battery life problematic.	SMS not well developed at present. Paging application not well supported.

In summary, paging retains a niche market within mobile communications of a core of around 800,000 business users for whom, for whatever reason, paging is the preferred means of communications. It is likely to retain this niche in the foreseeable future. It also has a role in the consumer market with CPP tariffs where current levels of growth are likely to continue, at least in the short term.

In order to enhance the services they offer to their customers, particularly the business customer, paging operators have been experimenting with information services such as the provision of stock market information to pagers. At present, the amount of information which can be sent is greatly limited by the low capacity of existing POCSAG paging systems. In the future, this capacity is expected to be enhanced by new high capacity paging technologies such as ERMES.

Paging operators have also noted that success in the consumer market depends on reducing prices. This might be best achieved through continued use of the existing networks which are already partly depreciated. Hence the operators can envisage a scenario where business users are catered for on the high capacity network, with a range of value added services, whilst household consumers are offered a relatively simple service at a low price on the existing POCSAG networks. The paging operators and manufacturers are still generally optimistic about the long-term prospects, expecting the circa 20% per annum growth to continue in the short to medium term. Some even expect to see penetration levels approaching 10% to be achieved.

An additional paging service may be developed alongside digital broadcasting. Legislation allows for up to 10% of the spectrum assigned to multiplex operators to be reserved for data transmission. This represents around an order of magnitude more spectrum than that assigned to the current paging operators. Given that little additional infrastructure would be required over and above that needed for digital broadcasting, then the investment required will be small compared to that required by a conventional paging operator. Hence if multiplex operators decided to offer a paging service, this could have a very significant effect on the overall paging market, reducing tariffs and increasing demand but having a negative financial effect on incumbents, especially those who have recently invested in new networks. There is no indication to date as to whether or when such a service might be offered.

If paging continues to grow at current rates and if two-way and voice paging services become popular, it seems likely that paging spectrum will be placed under increasing pressure in the future. Although the possible use of broadcasting spectrum for paging services could help meet this demand. Two-way paging will require additional spectrum for the return channel.

In the UK, licences have been issued which require the operators to use the ERMES standard. There is some concern amongst operators that the reduction in flexibility due to the requirement to use a particular standard will slow the market growth compared with what might otherwise be achieved. There is also concern that the standardisation of ERMES took so long that the product is already outdated, even before its first deployment in the UK (for example, it does not have capabilities such as voice paging). Some operators expressed concern to us that the Government might ask that POCSAG networks be withdrawn by the year 2004, when many are of the view that POCSAG will continue to be a useful platform for the consumer market.

This debate has been stimulated by the success of an alternative standard, termed Flex. This standard is proprietary to Motorola, however, a number of major manufacturers, including Philips, have agreed to make equipment conforming to the standard under licence. Flex has been extremely successful in the US and the Far East with over 10 million subscribers at present and over 25 million predicted by the end of 1998. There are about 300,000 ERMES subscribers at present. It appears that Flex is likely to be the most popular paging system on a worldwide basis. The reductions in price that then result from the greater economies of scale suggest that Flex equipment may be less expensive than ERMES and even POCSAG equipment. However, ERMES networks are currently under construction in some countries, products are available and hence it seems likely that ERMES will be available alongside Flex in many European countries.

This case study is not concerned with a debate over whether it is appropriate to mandate ERMES for future paging networks. However, the European Public Paging Association (EPPA) have started to lobby ETSI and the EC for freedom of choice in the implementation of paging standards in Europe. One operator noted the requirement that they compensate PBR users who suffer interference as a result of the introduction of ERMES could be easily avoided if they were allowed to use Flex in their existing POCSAG frequency allocation. Such freedom, it was claimed, would significantly improve the prospects for the paging market, providing much greater capacity than that offered by ERMES alone and providing operators with greater flexibility to develop differentiated services.

Paging is currently experiencing significant growth and is on the verge of deploying new technology which will increase the range of services offered. This growth is unlikely to be affected, at least in the short to medium term, by the advances in cellular technology. Paging penetration will increase significantly, although it is unlikely to approach the high levels experienced in the US and some Pacific Rim countries.

There are currently around 1,400 links in the 1.5GHz band owned by over 180 companies, with most companies having only around 4 links. The largest users are the utilities, ambulance services and some oil and gas companies requiring links to and between oil and gas platforms in the North Sea. Most links are low bandwidth and are used for control purposes, for example to shut down remote gas compressor stations.

The reason for studying this band is that it will be necessary for users to migrate from the 1.5GHz band (1450MHz-1474MHz paired with 1492MHz-1493MHz and 1507MHz-1530MHz) to the 1.4GHz band (1350MHz-1375MHz paired with 1492MHz-1517MHz), because the 1.5GHz band has been designated for digital audio broadcasting (DAB) on a European basis. The 1.4GHz band has been agreed on a pan-European basis as the low-frequency fixed links band. In this section we assess the cost of such a migration.

In principle, users could move to any fixed link band, to wired communications, or decide that communications is no longer required. In practice, most 1.5GHz links have a narrow bandwidth and the only fixed link band with a similar facility is 1.4GHz. Hence, for the vast majority of the links, a frequency other than 1.4GHz is not a viable option. Many of the links are long distance, or over terrain where it is difficult to lay a cable, and hence the use of a wired connection is not an economic alternative.

A potentially realistic alternative is the use of public data networks. Where the quantity of data transmitted is low, this can be an attractive option and some utilities already make use of public data networks in preference to fixed links. However, the coverage offered is relatively poor (around 90% of population) and hence many links to remote areas would not be possible over a public data network.

An alternative which may arise in future is the use of satellite-based transmission schemes. A number of store and forward low-orbit satellite systems (the so-called "little-LEOs") have been proposed which should provide adequate coverage and an economic alternative to many fixed links. However, the data rates available on such systems are unlikely to be suitable for most applications and there are also concerns about the delay imposed by store-and-forward satellite communications and the overall system availability and reliability. Our view is that such systems are likely to be unacceptable to most users of the 1.5GHz band.

We therefore conclude that most of the current users of the 1.5GHz band will choose to migrate to 1.4GHz as there do not seem to be any other viable alternatives.

Migration will inevitably require existing users to acquire new equipment. Transmitters and receivers will need to be replaced. However, it is quite possible that antennas, cables, masts and existing infrastructure can be reused due to the proximity of the two frequency bands. Since much of the cost of fixed links is the cost of site acquisition, provision of power and communications to the site, and the mast construction this will have the fortunate result that the migration costs will generally be significantly less than the initial construction costs. Based on the costs of typical 1.5GHz equipment, supplied to us by industry, we estimate that the new equipment required for both ends of a single link plus installation might cost Ł40,000 in total.

Since the cost of the 1.4GHz equipment is similar to that of the 1.5GHz equipment, then the true cost of migration to a user will be the depreciated value of their existing equipment. If their existing equipment is nearly new, the cost of migration will be almost the full value of their equipment, whereas if their equipment was due for replacement anyway then the cost of migration will be minimal.

Using these assumptions, a total cost of Ł28.8m can be calculated. This is a loss to the users who would otherwise not have to pay this sum and whose costs will be increased somewhat as a result. Since much of the fixed link equipment is likely to be manufactured overseas, it is a reasonable assumption that there will be relatively little benefit to other UK companies, from these expenditures apart from any additional installation/maintenance expenditures.

The reason for studying this fixed link band (7.425 - 7.9GHz) is that it is generally considered to be congested. The RA wishes to understand the impact on users and hence the broader economy as a result of this congestion.

It should be noted that the 7.5GHz band is not the only fixed link band that could be considered to be congested. A more comprehensive study of fixed link congestion might also examine other bands below 14GHz. We have been asked by the RA to focus on one band in order to understand the issues associated with congestion in more detail, rather than to propose any solutions to fixed link congestion in general.

There are currently around 870 links in this band, owned in the main by the broadcasters (BBC and NTL) comprising approximately 40%, the utilities (gas and electricity) at around 30%, the PCN operators (Orange and One2One) at around 15%, and fixed radio access (FRA) operators (Ionica) at around 3%. There are also a number of users with a small number of links.

One particular reason why the 7.5GHz band suffers from congestion is because part of the band is shared with military satellite services and there is a need to ensure that the main earth station site (which is unscreened) does not suffer from unacceptable levels of interference.

The congestion in this band has meant that on the main trunked routes (eg London to Bristol) user demands cannot be met and alternative routes or alternative frequencies have had to be adopted. In most cases, users choose to move to the 13 and 14GHz fixed link bands. One user commented that congestion at 7.5GHz has mainly been caused by short links, which could more suitably be accommodated in a higher band. The impact of congestion on key users is summarised in Table 4.5.

User	Requirement	Requirement Met	Alternatives	Effects
BBC	National network for programme distribution	Only in rural areas	Fibre	Minimal cost increase
NTL	National broadcast and telecoms network	In most, but not all cases	Fibre on some occasions, otherwise none	Unquantified loss of business
Utilities	General telecoms	Only in rural areas	Fibre	Minimal cost increase
Ionica	BS to switch links	Only in around 10% of cases	Leased lines	Significant cost increase
PCN operators	BSC to MSC links	In most cases	Higher frequency links	Slight cost increase

From Table 4.5 it is apparent that the band is congested in urban areas. Rules of thumb adopted by the users when understanding whether congestion is occurring include when no channels are available at given sites and when the rejection rate for new links reaches a significant level as perceived by the user. Although these rules of thumb are useful, it would be helpful to be able to quantify the extent of congestion. There are a number of possible measures, for example:

The usefulness of each of these measures is considered in more detail below. However, before proceeding to evaluate individual measures it is worth noting that all the measures rely on an understanding of the additional number of links which would use the band were there sufficient capacity (ie the excess demand).

The excess demand is often not clearly visible. In the 7.5GHz band there is evidence that some users did not apply for frequencies because they were sure that the request would not be granted. Hence, determining even the level of excess demand is highly problematic. The level of demand is also heavily influenced by the existing management technique. Alternative management techniques, such as the use of narrowband equipment or increases in the licence fees can be expected to change the supply and demand for the frequencies and so modify the extent of the excess demand.

This is the simplest measure of the cost of congestion to understand conceptually. For every user who would like access to the band but was denied access, the additional cost incurred by using an alternative means of communications is evaluated. The total of all these costs then forms the total cost of the congestion.

Despite these difficulties we attempt to estimate a value for this measure below.

Another way to measure the cost of congestion would be to assess the cost to users of adopting equipment that was sufficiently more efficient than existing equipment to reduce the excess demand to zero.

All these difficulties are severe and prevent further evaluation in this report of this measure in the case of the 7.5GHz band.

A further way of evaluating the cost of congestion would be to understand the costs involved if all those who required to use the band were allowed to do so, generating higher levels of interference to each other than at present. This would result in a lower utility from each link and hence a lower value to the user of having the link.

Amongst all these problems, it is the difficulty in understanding how the value of the link to the user falls with increasing interference which prevents any further evaluation of this cost.

As explained above, one approximation to the economic cost of congestion in the 7.5GHz band is given by the number of links which could not be deployed due to congestion multiplied by the additional cost suffered as a result of having to use alternative means of communications. Both of these figures are difficult to determine accurately. An estimate of the number of links which could not be deployed by the main users of the band is given in Table 4.6. Also shown in this table is an estimate of the average additional cost for each user, which differs between users according to their circumstances. These costs have been derived both from discussions with the users and our own estimates which show that the additional costs are:

The lack of quantitative information in this table shows the difficulties involved. In particular, the BBC and the utilities had not quantified the additional cost and since details of their agreement with Energis are confidential it is not possible to do this. NTL were unable to quantify the number of links they would have liked to install. With Ionica, the number of links was judged to be confidential and hence it was necessary to estimate this, although the additional cost per link can be estimated with some confidence. The cellular operators had not kept records of exact number of links refused and hence could only make very approximate estimates.

Given the difficulties and approximations in calculating this figure, only tentative conclusions should be drawn from it. A first glance at the table would suggest that the impact of congestion in this band is not severe with most organisations only suffering minimal costs as a result. However it is important to note that Ł31m is a lower bound and the actual figure may be significantly higher than this since there may be other organisations who would like to use the spectrum but have been denied access (and existing users who do not apply for spectrum because they believe they will not get it). It appears that established users of the band are not particularly affected by congestion but new telecommunications operators could face additional start-up costs. Unfortunately our assessment of the impact of congestion has been severely constrained by the lack of factual information that we have been given by users/potential users of the 7.5GHz band.

Given the costs of congestion in the 7.5 GHz band are significant, the issue arises as to whether it would be worthwhile for some fixed link users to compensate the military for reducing the impact of its unscreened satellite earth station on the availability of frequencies in the band. A cost/benefit appraisal of the options could well be a worthwhile exercise.

We were asked by the RA to investigate what particular advantages are provided to the UK economy by the location of Inmarsat in the UK.

Inmarsat is an international treaty organisation which was established in 1979 to provide global mobile satellite communications through a geostationary satellite system. The organisation is based in the UK, as a consequence of its origins as a provider of services solely to the maritime community and hence the advantage of being located in a maritime nation. The choice of location was also influenced by the fact that the UK (BT) is a major signatory to Inmarsat.

Inmarsat provides direct value to the UK in terms of employment and expenditure on items produced in the UK. The Inmarsat accounts for 1995 show expenditure on staff costs, accommodations costs and office services totalling $59.5m (around Ł36m). Not all these expenditures will occur in the UK, but in the absence of more detailed information we have assumed that most are UK-related.

Inmarsat also provides indirect benefits by encouraging UK-based manufacturing and service industries. Inmarsat was unable to provide us with information as to the size of these benefits. However, one example of such a benefit is that of the location of ICO Global Communications Limited, in which Inmarsat is a shareholder, in the UK. ICO is a UK private company engaged in the design, development, construction and financing of a personal mobile communications system. It is likely that ICO is based in the UK largely because of the system’s origins in Inmarsat. The company’s operating expenses for 1995 were $26m (around Ł16m), however it has not been possible to ascertain how much of this has contributed to the UK economy.

	Paging	PBR	PAMR	GSM	PCN
Bandwidth*	25kHz	12.5kHz	12.5kHz	25kHz	25kHz
Subs/MHz	500,000	17,000	5,000	69,000	13,000
Traffic/MHz/day (Mbits)	110	16,000	24,000	61,100	29,700
Re-use*	25kHz	150kHz	150kHz	225kHz	225kHz

User	Number of links not deployed	Additional through-life cost / link	Total cost
BBC	Not known	Minimal	Minimal
NTL	Not known	<Ł150,000	Not known
Ionica	~270	~Ł100,000	Ł27m
Utilities	0	Minimal	Minimal
Cellular operators	<10	Not known	~Ł4m
Total			>Ł31m

Economic Impact of Radio '95