Impact of Satellites on UMTS Network

Impact of Satellites on UMTS Network R J Finean, D Polymeros, A El-Hoiydi, F da Costa, M Dinis, A Saïdi, B Vazvan BT 1, OTE, Ascom, IT, IT, Alcatel CRC, Nokia Abstract: UMTS will appeal to the existing cordless, paging and cellular markets and to the emerging satellite personal communications market by allowing standard delivery of a diverse range of communications services to people, no matter where they are. Satellite communications will feature as the preferred mode of access to rural and remote regions as well as being a means to rapidly deploy UMTS service at the initial commercial roll-out of UMTS networks. Race Monet, in conjunction with Race Saint, have developed a mobile network architecture capable of providing UMTS services through a wide variety of satellite networks in the same way as it provides service to a wide variety of terrestrial radio environments. This paper presents the features of the UMTS system specification that were added specifically to support satellite access networks. These include enhancements to the functional entities which handle location update, domain update and handover. A description of the assumptions that have been made and an indication of the performance of the system specification when satellite access is used are presented. Approach to Defining Network Functions for Satellites The Race Monet project had been working on network models for three years before satellite aspects were considered in detail. The satellite engineers approach was to make minimal changes to the terrestrial derived model and to keep the network model generic, to allow participation in UMTS of a number of different satellite networks. For satellite use, network design needs to account for the possibility that the satellites (acting as cell site antennas) might be moving overhead very fast relative to the mobile terminals (MTs) and to the Fixed Earth Stations (FESs) which act as the point of connection from the mobiles to the core UMTS network. Another consideration is that satellite bandwidth is expensive and that its use for signalling needs to be minimised. Few changes to the terrestrial model have been made, all of which have been confined to location update and to handover. Location Area and its Update A location area is formed by an FES s instantaneous coverage. Using the same functional model used for terrestrial cellular networks, an MT will location update only if it looses the FES s location area broadcast channel. From the network viewpoint, the location of the MT is somewhere within reach of the FES. The FES (either on its own or with the help of the MT) may provide a way of intelligently reducing the area over which it pages in the event of an incoming call. An assumption is that each FES will be designed to cover a particular geographic area. To maintain coverage of this area whilst allowing for satellite orbital motion an FES may need to use a number of different satellites and it may need to share satellites with other FESs at certain times. Because of the dynamics of a satellite network, the edges of an FES s coverage are intermittently covered by this FES. However, for each FES it is possible to define a guaranteed coverage area (GCA) which is the geographic area over which the FES is designed to provide service 1% of the time. The FES will be programmed to always cover the GCA but minimise the transmission of its location area identity outside the GCA. In a satellite system designed to provide coverage throughout a region, these FES 1 BT Laboratories, MLB4/67 Martlesham Heath, Ipswich IP5 7RE, UK. E-mail: bob@garden.bt.co.uk

GCAs will overlap in places and combine to cover the region with no gaps. If an MT is in overlapping FES coverage and location updates to an FES only intermittently covering its location (because the MT is not in the FES s GCA), it will loose that FES s location area broadcast after a while and be forced to location update to another FES which covers its location properly. Examples of Guaranteed Coverage Areas A GCA can be characterised by γ MT, the minimum satellite elevation angle that is tolerated by an MT Inmarsat P21 Globalstar and by the type of desired coverage (single or multiple satellites). Inmarsat P21 provides generous multiple coverage at all latitudes, as does Globalstar at temperate - - - - - - latitudes. The figures on the left show the largest possible GCA for each, from an FES at N, E using all satellites at elevations above 5. The outer contour is the GCA with at least a single satellite at γ MT 2. The inner contour is the GCA for at least two satellites visible, both with γ MT 1. The smallest diameter of the single and diverse GCAs are for Globalstar Globalstar 34 km and 22 km, respectively, and for Inmarsat km and 33 km. Closer to the equator, Globalstar satellites are more sparse and the requirements for a GCA must be reduced for these latitudes to guarantee only single satellite coverage with γ MT 1. The figure on the right shows such a GCA for an FES at N, E. Iridium provides single satellite coverage with very little overlap at the equator but its use of inter-satellite links allows an FES to guarantee coverage of an area as large as desired. Paging Implementation Options - - - As in the terrestrial segment, any incoming call will be routed to the FES which can guarantee that the MT, if it is working, is within the area the FES is covering with its broadcast channels. With the FES at its simplest, the FES would then transmit a paging message for the mobile through every spot beam which it is using to cover its GCA. If incoming calls occur infrequently, this may be acceptable but otherwise paging through this many spot beams is considered a waste of power and spectral resources. Satellite network designers could use a number of techniques to reduce the number of spot beams paged, some examples of which are presented below. Using Multiple Location Areas per FES As shown above, the GCA of an FES can be very big, so it may be convenient for the FES operator to split the area into two or more location areas, each with a distinct broadcast location area. As with the GCA, these location areas would be geographically fixed. For example, an FES covering southern Europe, the Mediterranean and the Middle East might split the location area along a border which is seldom crossed, say the middle of the Mediterranean sea. This then reduces the maximum area over which paging is necessary whilst increasing the location update signalling traffic only marginally.

Paging Areas Smaller than the Location Area - Intelligent Paging Other approaches use information about the MT s position to avoid paging throughout a location area and only page through spot beams where it is likely that the MT is. This is intelligent paging [1]. In each of the following cases the FES, possibly with help from the MT, uses additional information in order to page the MT in a paging area that is smaller than the location area. The network beyond the FES is not affected by this extra intelligence. A) Using Location Update Spot Beam Position A first approach to intelligent paging could be the FES identifying and recording the instantaneous size, shape and location (latitude, longitude) of the spot beam in which the MT last made contact (for location update, call set-up or any other reason), with a time-stamp. In the event of an incoming call, the FES would page only those spot beams required to completely cover the recorded area in the first instance. In case the mobile does not respond the paging is repeated over a wider area, depending on the age of the time-stamp and on the MT mobility profile. B) Using a Terminal Position Fix If an MT is capable of making the necessary measurements and calculations to fix its own position then the FES could record the measured position of the MT, which would be more accurate than a spot beam area. The location update message from the MT to the FES would be modified to include the latitude and longitude of the MT and an uncertainty radius that determines the circular area where the terminal can be found at any time. The MT would continuously monitor its own position and if it moves outside of its uncertainty area it would perform a position update. If the MT is still receiving the location area broadcasts of the FES then position update could remain local to the MT and FES, since from the network point of view the MT is still contactable through the same FES as before. However, if at any time the MT lost the location area broadcast of the FES then it would search for and location update to a new FES, invoking the full location update procedure defined by Race Monet. On an incoming call, the FES only needs to page the spot beams that cover the MT s declared uncertainty area (or at least the part of it within the FES s location area). Depending on the mobility of the MT and its users incoming and outgoing call rates, the MT might vary the uncertainty area radius to minimise either the paging area or the number of position updates in an attempt to minimise the total spectrum and power resource consumed by this signalling. Globally one can show [2] that the paging signalling requirement rises linearly as the uncertainty radius increases whilst the position update signalling decreases rapidly. Disadvantages of this approach are the extra complexity in the MTs and the additional space resources needed to implement position fixing. A terminal capable of position fixing will be more expensive to produce than one that is not but position fixing might be useful as a UMTS service for other purposes 2. Position fixing signals could either be supplied by a third party (e.g. the US Navy s GPS system) or extra bandwidth on UMTS satellite broadcast channels could be reserved for the timing signal and satellite almanac details. 2 Note that the accuracy required for position fixing as a UMTS service might be different to that required for location management. The accuracy of a positioning system like GPS would be expensive to match.

C) Using Dual Satellite Coverage Position Fix If an MT is covered by two or more satellites (both in use by the FES) when it makes its location update then the FES could perform position fixing ranging measurements for itself. Based on the relative measured delay and Doppler frequency shift via the different satellites, the terminal position could be calculated and stored along with an uncertainty area and time-stamp. As in approach A, this position would not be updated unless a call was set-up or the MT lost the FES location area broadcast. On an incoming call, the FES would page an area covering the uncertainty area or a wider area, depending on the age of the position time-stamp. Unlike approach B, this FES intelligence does not require any extra position fixing features in the MT. A periodical position update could be implemented, either MT or FES initiated, to reduce uncertainty of old location updates. Comparison of Some of the Intelligent Paging Options Alternative A demands very little location update signalling but the approach in B is more economic with paging signalling because the position is known more accurately. Based on a simple mobility and traffic model, one can analytically derive a measure of the total signalling load for each scheme. The two fundamental parameters of the model are the incoming call rate R in and the movement speed v of the users. The conclusion is that approach A is the best when v is large and R in low (high mobility and few incoming calls), i.e. when the location update signalling load is higher than the paging signalling load. At the time of writing, it is difficult to predict whether the UMTS market will favour one approach or the other. Handover Network Functions In UMTS the most difficult satellite handovers will be those to and from other networks (including terrestrial) and those between different FESs in the same network. The differences in radio path may cause short interruptions on handover but Race Monet decided it is necessary for the network to be able to support these types of handover, since any handover is preferable to redialling and setting up a new call. As with location update, satellite considerations have not resulted in changes to the structure of the Monet handover functional model but the roles of some of its entities have been expanded. These modifications are mostly related to the monitoring of the satellite link and to the access rights and preferences of subscribers to the two segments (terrestrial and satellite). Monitoring of Satellite Links Handover Criteria Adjustment (HCA), Measurement Function (MEF) and Target Cells and Connections - User (TCCU) functional entities have all been expanded to include satellite connections. The MEF gathers measurements concerning a specific active link. The HCA sets and updates the handover criteria, which the Handover Initiation (HI) entity will act on, according to network management and resource control instructions. Because the criteria and thresholds for determining Information Gathering Phase Decision Phase Execution Phase MEF TCCU HUPU SHRU connection quality will be different for different types of satellite and terrestrial connections, these entities need to be aware of the different types of networks to which the MT has access. BC r1 HUPN TCCN r SBC r2 CMC r16 r9 r1 r14 r17 r18 r3 CIC HI HD r8 HOC r4 r7 r11 r12 r13 r19 r2 AAC r21 r5 RRT SHRN CPT Monet Handover Functional Model r22 r23 HCA HC r6 Cold Cnew r24

The TCCU provides a list of optimum cells or spot beams to the HI for a specific terminal. It needs to gather the information (e.g. quality and spare capacity) provided by all available cells and spot beams, irrespective of which network the MT is communicating with at the time. For satellite-toterrestrial handovers this is very important because the Target Cells and Connections - Network (TCCN) functional entity in the network may not be able to identify an appropriate target cell to handover to in the terrestrial segment. The TCCU must be able to support this function. Customer Access Rights and Preferences The Special Handover Request (SHRU and SHRN) and Handover User Profile (HUPU and HUPN) provide the means and information for handovers based on a customer s preferences and access rights. The SHRU provides the opportunity for the user or their terminal to ask for a handover in order to select a preferred coverage domain and the SHRN allows the network to initiate a handover for specific reasons (e.g. traffic management). The HUP contains a subset of the user s profile, namely all the information relevant to the handover process (bandwidth requirements, quality requirements, priorities, environment selection etc.). In the integrated cellular and satellite UMTS network it will contain information about the access rights and preferences of the customer. For example, a customer may subscribe only for cellular services - in this case the inter-segment handover cannot take place. Alternatively, a customer may want the terminal to ask its user before it hands over to the satellite segment. Finally, the Handover Decision (HD) entity receives a handover execution request from the HI, determines the location of the handover control point and takes the final decision about handover execution. This final decision is based on an algorithm responsible for assessing the various data available. The way this algorithm makes a decision whether or not to request a handover must consider the additional information from the satellite requirements, as described above. Conclusion With the revisions described in this paper incorporated, the Monet network functional model will enable the design of UMTS networks just as capable of supporting mobility in satellite cells as it is in terrestrial macro-cells and micro-cells. Acknowledgements The authors wish to thank all the other partners in Race Monet work package RAS2 (satellite aspects) and in Race Saint work package A41 and the Race Monet project management team. References [1] PTT Nederland contribution to Race Monet: Location Areas, Paging Areas and the Network Architecture. R266/PTTNL/MF1/DS/P/1/b1. [2] C Cullen, A Sammut, R Tafazolli, B Evans: Network and Signalling Aspects of a Satellite Personal Communications Network. 1 st European Workshop on Mobile/Personal Satcomms (EMPS 94), ESA/ESRIN, Frascati, October 1994.