Aspects for the integration of ad-hoc and cellular networks

Transcription

1 3 rd Scandinavian Workshop on Wireless Ad-hoc Networks, Stockholm, May 6-7 th 2003 Aspects for the integration of ad-hoc and cellular networks Gabriel Cristache, Klaus David, Matthias Hildebrand University of Kassel Department of Electrical Engineering / Computer Science Chair for Communication Technology (ComTec) D Kassel, Germany Tel. +49 (561) , gabriel@comtec.e-technik.uni-kassel.de, david@uni-kassel.de José Díaz, Rolf Sigle Alcatel SEL AG Research and Innovation D Stuttgart, Germany ABSTRACT Ad hoc networks are becoming more and more successful in the marketplace as indicated by the increasing usage of Bluetooth and W-LAN networks. This success is due to several reasons, including their lack of infrastructure requirements as well as their suitability for several application scenarios like the communication between laptops and/or handheld devices. Despite all the successes so far and the promising future of ad hoc networks, Internet connectivity will still be very important, and could quite likely be supported by cellular networks like GSM/GPRS or UMTS. Another possibility is to use ad-hoc networks as an extension of the cellular air-interface, leading to improvements in cellular capacity and coverage. Due to the reduced coverage range of Bluetooth, WLAN seems to be better suited for this purpose, and hence, different interworking solutions between UMTS system and WLAN are presented in this paper. Initially, current approaches, based on the coupling of WLAN access points into the UMTS infrastructure, are briefly introduced. However, the paper is mainly focused on the interworking of cellular and WLAN in ad-hoc mode, i.e. without using WLAN access points. The results show a good potential for cellular capacity increase based on ad-hoc extensions. I. INTRODUCTION The most common and widely used wireless access systems today are cellular radio systems, based on GSM, GPRS and UMTS standards. Cellular systems and WLANs should be considered as complementary systems: cellular systems could provide universal coverage and high mobility support while WLANs will be applied in hot spot areas offering high data rates. This observation motivates the investigation of possible combinations between these two technologies and their potential benefits. One possible approach is the coupling of WLANs like IEEE a/b with cellular networks like UMTS being discussed in Section II. Another approach is the extension of the cellular coverage by means of ad-hoc networks, which is presented in Section III. In Section IV the capacity gain of the integrated ad-hoc and cellular system is estimated and in Section V a summary is given. II. COUPLING BETWEEN UMTS AND WLAN The coupling between WLAN and UMTS can be implemented in different ways. 3GPP, has defined six scenarios for the coupling, each of them enabling a particular feature and requiring an increasing level of integration for the interworking [1, 2]. The goal of the coupling is on the one hand to combine the wide-area coverage of UMTS, with its associated roaming and mobility properties and, on the other hand, to gain additional throughput and capacity by using WLAN in hotspots. A general design criterion is to reuse existing mechanisms and functions, e.g. UMTS subscriber management mechanism, billing functions, UMTS authentication, and security functions. The six scenarios defined by 3GPP are focused on the type and quality of the service offered to the user. However, from an architectural point of view, these scenarios can be reduced to four levels of coupling, which are shortly discussed in this section. The first level is open coupling, presented in Figure 1. In this case UMTS and WLAN make use of two separated access and transport networks and billing is common although using different authentication mechanisms. The next level is the so-called loose coupling, shown in Figure 2, which enables the use of common authentication mechanisms by providing a link between the authentication, authorization and accounting (AAA) server in the WLAN subsystem and the Home Location Register (HLR) in the UMTS subsystem, which are still kept separate. 1

2 Figure 1. Open Coupling UMTS-WLAN Interworking Unit might be added, which is discussed in [2]. With tight coupling the handover between WLAN and cellular subsystems could be supported. Finally, with very tight coupling (Figure 4) the WLAN access point is connected to the RNC and, thus, it becomes an integral part of the UMTS Terrestrial Radio Access Network (UTRAN). Very tight coupling provides the possibility to offer a homogenous service to the subscriber and to perform inter-system Radio Resource Management (RRM). The complexity of implementation increases with the level of interworking from the open coupling solution (the simplest one) up to very tight interworking. A higher degree of interworking involves a higher impact on UMTS network elements, which have to provide the necessary capacity with respect to processing power and interface capabilities. However, the different approaches presented here for WLAN coupling are based in the WLAN infrastructure mode (using access points). Another option for the integration of WLAN and UMTS, based on the use of the WLAN ad-hoc mode, is discussed in the next section. III. AD-HOC EXTENSION OF CELLULAR Figure 2. Loose coupling UMTS-WLAN In this section, a possible ad-hoc extension of the UMTS cellular system is presented. In this paper, the one-hop case is considered for the ad-hoc extension, see Figure 5. The main idea is that it is possible for the mobile terminal (termed mobile client) to communicate Figure 3. Tight coupling UMTS-WLAN Figure 5. An example of the ad-hoc extension Figure 4. Very tight coupling UMTS-WLAN With tight coupling, the WLAN Access Point (AP) is connected like a Radio Network Controller (RNC) to the Serving GPRS Support Node (SGSN) in the Core Network (CN). This principle is shown in Figure 3. An with the base station not only directly but also using another mobile in between that performs relaying (and hence termed mobile relay). This also means that the communication between the mobile terminal and the base station is indirect, via two different links. There are a number of possibilities to facilitate the separation of cellular and ad-hoc links. Possible solutions can be: Different air-interfaces (implying also different frequency spectra) Identical air-interfaces in different frequency spectra Identical air-interfaces and frequency spectrum. 2

3 In this paper, the first approach is considered in which different air interfaces and different frequency spectra are used. In other words, the first link (between the base station and the mobile relay) is cellular while the second link (between the mobile relay and the mobile client) is ad-hoc, both being orthogonal, i.e. causing no interference on each other. As a result, potential benefits can be envisaged for cellular and also for ad-hoc networks. On the one hand, new services can be made available in the ad-hoc networks. On the other hand, the expected benefits for the UMTS cellular network are threefold: Cellular coverage may be extended by means of adhoc relaying The intra-cell and extra-cell interference may be reduced by using relaying, what may result in an increase of capacity Load balancing between different cells can be provided by means of ad-hoc relaying In a cellular communication system, terminals that are far away from the BS (base Station) need to use an overproportional large amount of transmission power (due to the non linear decrease of the received power with the distance) [3]. With a code division multiple access (CDMA) scheme from UMTS, the higher the power used for a particular terminal, the higher the interference experienced by the rest of the users, either in the same cell or in neighboring cells (since the same frequency band is used and the channelization codes are not perfectly orthogonal in practice). In the relay communication case, i.e. when using an intermediate mobile in between, the transmission power for the cellular link can be reduced. This fact has as a direct consequence a reduction of interference. Based on the observations above, an increase in capacity of the system is expected. For quantification, an investigation on capacity improvement based on simulation is presented in the next sections. IV. CAPACITY ESTIMATION A first estimation of the capacity improvement of the UMTS system using ad-hoc extensions can be made using an event-driven simulator for UMTS [4]. In a first step the capacity and coverage is simulated for the UMTS system. In a second step, using these results a first statement about possible capacity improvements in the integrated system (ad-hoc plus cellular) can be derived. The simulation environment for the event driven simulator is presented in subsection A and the results of the simulation are presented in subsection B. The achievable downlink capacity gain that results from the use of mobiles as relays for other terminals is evaluated in subsection C. A Simulation Environment For this investigation a scenario with 42 cells is considered (Figure 6). The cells are arranged in seven columns and six rows and a hexagonal pattern is used for the cell shapes. Each cell has a base station using an omni-directional antenna at its center. The nominal cell radius is defined as 800 m. Only the situation in the central cell is under investigation and the load of all other cells is set to maximum transmission power thus creating maximum interference. The mobile stations are considered as stationary, but they are placed randomly for each call. For the sake of simplicity the investigation considers the downlink only, which is more critical due to the asymmetric traffic generated by packet-based applications such as web browsing. Figure 6. Simulation area. The central cell, shown in black color, is under The calculation of the individual transmission power necessary to satisfy each user in a CDMA based system was described in [4]. The main idea is that for each link a certain SIR target value is defined based on the Quality of Service required. The system tries to keep the current SIR of a link close to its target value by using a tight power control. The total transmission power of each base station is set to 20 W. The path loss model is based on the urban macro-cell model proposed in [5] and the orthogonality factor is 0.4. The event driven simulator creates calls with random interarrival times based on an exponential distribution. For each call a mobile terminal is placed randomly within the investigation area based on a uniform distribution. The assignment of a call to a particular cell is based on the SIR of the pilot signals. Since the current transmission power of the surrounding cells is set to its maximum allowed value, these cells seem to be constantly fully loaded. Hence, all arriving calls are forced to connect to the central base station thus creating the highest possible load as well as extending the central cell coverage to the maximum area that still can be supported in the given situation. During a simulation run, the number of mobile terminals and the distance to the furthermost mobile terminal are constantly monitored. Three different user data rates have been evaluated: 8 kbps, 16 kbps and 32 kbps, as depicted in Figure 7 [6]. 3

4 B Simulation Results The starting point for the estimation of the capacity for the system in which some mobiles are used as relays for other terminals is based on the observation of Figure 7, where the maximum cell capacity decreases as the cell radius increases. Figure 7. Coverage capacity simulation results For instance, for 8 kbps (voice service) about 85 users could be supported within 650 m of coverage, whereas this number falls down to around 60 users at 800 m. This result is normal in a typical CDMA environment where the cell is breathing [6]. This effect describes the variance of the cell coverage due to the change of load, where the load is defined by several parameters like user data rate, number of current users, required quality of service, and others. C Interpretation of the results In the previous example, the numbers of 85 and 60 users respectively would suggest a 1.41 times capacity improvement at 650 m as compared with 800 m cell range. For an optimum distribution of mobile relays, the adhoc extension of the cellular coverage can be used to reduce the radius (by 150 m in the example) that has to be covered by the cellular system thus increasing the capacity. Here the difference of 150 m can be covered using e.g. a WLAN ad-hoc extension. The idea is illustrated in Figure 5. As shown in the figure, the cellular base station covers only the inner part of the cell with a radius r smaller than the nominal cell radius R. The area of the ring between this inner circle and the cell border is covered by mobile relays using the ad-hoc air interface. Since the capacity C of a cell depends on the area to be covered we can define C as a function of the radius: C(r) > C(R) with r < R Hence, the relative capacity gain g can be defined as: C( r) C( R) g = C( R) Using Figure 7, capacity values (in terms of number of users) can be achieved for different values of the cell radius. For example, when 8 kbps connections are considered, 60 users can be served in a radius of 800m, whereas 85 users can be served within 650m of the base station, i.e. C(800 m) = 60 users and C(650 m) = 85 users. By supporting all mobile terminals at a distance between 650 m and 800 m using mobile relays based on WLAN, the UMTS coverage required the whole area is reduced to 650 m, thus increasing the number of users that can be supported at a distance of 800m from the UMTS cell by about 41 % (see Table 1). It is assumed that a sufficient number of mobile terminals are located within the inner circle and can act as mobile relays, providing WLAN coverage to the mobile terminals within the outer ring. This assumption can be justified because in a real cellular system, idle terminals which could potentially act as relays are always much more numerous than connected terminals. Mobile terminals in the outer ring can only act as mobile clients via WLAN, whereas terminals inside the inner circle can act as mobile relays but are not allowed to be served by other mobile relays.. For the actual assignment of mobile relays and mobile clients a selection algorithm has to be applied, which is out of the scope of this paper. The 41 % increase in capacity, if mobile relays are used, can be explained by the fact that it allows a reduction of the downlink transmission power on the UMTS air interface. Another reason is that extra spectrum from the ad-hoc extension has been used. The observation has been extended to other distances and the results are summarized in Table m 600 m % 500 m 650 m % 550 m 700 m - 62% 150% 600 m 750 m 42% 66% 300% 650 m 800 m 41% 80% - Table 1. Capacity improvement using ad-hoc extension (150 m WLAN) The evaluation in Table 1 is limited to 800 m, because this is the nominal cell radius of the system considered in the simulation. The ad-hoc extension of cellular could be additionally exploited to establish load balancing between adjacent cells [7], using inter-system radio resource management strategies which are out of the scope of this paper. To study the impact of different user data rates, the same procedure described before was applied. The results are also summarized in Table 1. For the same number of users the cell coverage decreases when increasing user data rate [6]. This has also an impact on the capacity improvement. As it can be observed in the table, the relative capacity improvement is increasing with the data rate. For 32 kbps up to 300 % capacity improvement could be achieved. The investigation has been extended for WLAN coverages to 100 m and 50 m respectively. The results are summarized in Tables 2 and 3. 4

5 As expected reducing the distance of the ad-hoc extension reduces the capacity improvement. 450 m 550 m % 500 m 600 m % 550 m 650 m - 44% 66% 600 m 700 m 33% 25% 100% 650 m 750 m 21% 50% 200% 700 m 800 m 25% 60% - Table 2. Capacity improvement using ad-hoc extension (100 m WLAN) 450 m 500 m % 500 m 550 m % 550 m 600 m - 30% 25% 600 m 650 m 17% 11% 33% 650 m 700 m 13% 12.5% 50% 700 m 750 m 7% 33% 100% 750 m 800 m 16% 20% - Table 3. Capacity improvement using ad-hoc extension (50 m WLAN) As a conclusion of the above investigation, the capacity improvement achievable by ad-hoc extension of cellular is dependent on the average range of coverage of the adhoc extension and on the user data rate. One explanation for the latter factor is that the absolute values for coverage as well as for user number decrease with increasing data rates (since they require a higher transmission power to maintain the energy per bit ratio, hence causing a higher interference). Since the curves in Figure 7 are roughly parallel the absolute capacity improvement in terms of number of users is similar for all data rates. Therefore, the higher the considered data rates, the better relative improvement in capacity. The achievable improvement depends also on the configuration considered (omni-directional antennas, 20 W maximum transmission power, 800m nominal cell radius). For other configurations the results will vary [6] but the basic behavior is expected to be the same. V. SUMMARY This paper introduced different interworking solutions between UMTS and WLAN. First, existing coupling alternatives for the integration of WLAN access points in the UMTS architecture have been discussed. The options presented differ in complexity of protocol integration and e.g. whether common RRM is possible or not. However, an additional alternative for the integration is using WLAN in ad-hoc mode as an extension of the UMTS airinterface. Based on simulations of the UMTS system, investigation results on possible capacity increase were presented. The explanation is twofold: Reduction of the UMTS transmission power that is required for distant users due to the use of mobile relays, which results in reduced distances for the cellular communication link. The integration of the ad-hoc relays adds resources to the system, i.e. additional spectrum has been utilized. However, the absolute values of the capacity increase should be considered carefully, since they are very dependent on the selected scenario (e.g. WLAN range, service, etc.). In future investigations, signaling scenarios to enable ad-hoc extensions will be considered, which could decrease the theoretical capacity gain. ACKNOWLEDGEMENTS We would like to acknowledge the support of Alcatel Research and Innovation (Stuttgart) and partial funding of the German Bundesministerium für Bildung und Forschung (BMBF) in the framework of the IPonAir project. REFERENCES [1] 3GPP TS : Feasibility study to Wireless Local Area (WLAN) interworking, V1.1.0 ( ) [2] M. Litzenburger, H. Bakker, S. Kaminski, K. Keil, Very Tight Coupling of Wireless LAN and UMTS Networks: A Technical Challenge and an Opportunity for Mobile Operators, Personal Wireless Communications Conference 2002 (PWC 2002), Oct , Singapore [3] H. Karl, S. Mengesha: Analysing Capacity Improvements in Wireless Networks by Relaying, Proc. of IEEE Intl. Conf. on Wireless LANs and Home Networks, pp , Singapore, December [4] M. Hildebrand, G. Piao, K. David, G. Cristache and F. Fechter: "Efficient Estimation of Transmission Power applied to The Simulation of the Cell Breathing Effect in CDMA-based Wireless Systems", IASTED International Conference "Applied Modelling and Simulation (AMS 2002)", November 4-6, 2002, Cambridge, USA [5] 3GPP TR : RF System Scenarios, V4.1.0 ( ) [6] M. Hildebrand, G. Piao, K. David, G. Cristache, F. Fechter, R. Sigle and A. Warich: Investigation of Cell Breathing Effect in CDMA-based Cellular Systems, accepted for publication at IST Mobile & Wireless Communications Summit 2003, Aveiro, Portugal. [7] Jie (Jay) Zhou and Yang Richard Yang. Parcels: Pervasive Ad-hoc Relaying for Cellular Systems, In Proceedings of Med-Hoc-Net, Sardegna, Italy, Sep