Analysis of the inter-system interference with respect to the required minimum frequency separation Peter Seidenberg, Matthias Lott Abstract This paper deals with the inter-system interference of mobile communication systems that are located in adjacent frequency bands. To reduce these interferences, the systems usually are separated by an unused frequency band. In this paper we present a method to evaluate the required minimum frequency separation (MFS) of coexisting mobile communication systems. Based on this comprehensive evaluation concept, this paper discusses the ability of cellular systems to bypass critical interference situations by applying power control and handover. onsidering these protocols the required minimum frequency separation decreases dramatically. Especially the MFS between GSM and TETRA as well as between UMTS (TDD-mode) and DET is studied. It is shown that most of the band that hasn t been assigned so far can be used to carry traffic. I. INTRODUTION In respect to an efficient allocation of frequency spectrum to mobile communication providers the characteristics of the planned resp. existing mobile communication systems have to be considered to ensure the simultaneous, undisturbed service of mobile radio systems which are operating in adjacent frequency bands. Especially, this task will become important when planning future systems like the Universal Mobile ommunication System (UMTS) that is based on different mobile system standards that have to reside in an appropriately chosen frequency band. To avoid harmful interferences of coexisting mobile communication systems an appropriate minimum frequency separation is introduced between systems in adjacent frequency bands. A method for MFS evaluation has been presented in []. This method takes into account the interferences due to simultaneous operation of the systems as a function of the transmitted power, attenuation and carrier frequency. As 2nd and 3rd generation s mobile communication systems are intelligent enough to bypass critical interference situations, this paper shows how this network intelligence can be considered in calculating the MFS. It is shown that the ability to perform power control and handover allows to use the so far unused MFS band to carry traffic. The paper is organized as follows. First the concept of evaluating the MFS as presented in [] and [2] is summarized. Section III describes a model for considering power control protocols in the calculation. A Model for handover is presented in section IV. As an example, in section IV-A the MFS is calculated for the systems TETRA (Trans European Trunked RAdio) and GSM (Global System for Mobile ommunications). Section IV-D gives an idea of the capacity gain that can be reached Aachen University of Technology, hair for ommunication Networks, Prof. Dr. Ing. Bernhard Walke, Kopernikusstr. 6, D 5274 Aachen, Germany. E mail: flott,psg,smdg@comnets.rwth aachen.de. by using potential MFS. At last, the MFS between the UMTS (TDD-mode) and the DET band is investigated in section V. II. THE MFS ALULATION ONEPT The minimum frequency separation (MFS) is the unused frequency band between two different radio systems intended to decrease the possibility of mutual interference. [] presents a method to evaluate the MFS considering emissions of interfering stations and the impact of these emissions at the perturbed receiver. The calculation method is based on a Monte arlo (M) simulation technique. To describe the general situation that is investigated an example interference scenario is shown in Fig.. Two different cellular systems overlayed in the space domain are working in adjacent frequency bands. The spectral characteristics of the in- Fig.. Example interference scenario terferences are described with the help of masks. The mask for the interfering transmitter represents the maximum permissible unwanted emission levels as a function of frequency. These values can be taken from the relevant standard specifications. The receiver characteristic is also represented by a mask that can be found by transferring the interference rejection mechanisms defined in the standards to an equivalent carrier-to-interference (/I) ratio []. Depending on a number of parameters like user densities, antenna heights, coverage radii, transmitter power etc. that determine an interference situation depicted in Fig. the interference at a perturbed receiver station is evaluated with the help of the masks. Each M calculation cycle starts with the positioning of the receiver station by means of an appropriate distribution function for the user path. This receiver becomes the victim receiver due to interferences of the interferer stations. Depending on the distance between transmitter and receiver the propagation loss is calculated using a hybrid Hata-Okumura, ost-23- Walfisch-Ikegami model []. The same is done for the links in
the interfering system. All present =I values, whether they are measured in the receiver band or outside this band, are calculated using the transmitter masks and are compared with the respective required carrier-to-interference ratios =I req defined by the receiver masks at the respective frequency f i. The minimum difference between the present and required =I ratio min f i I (f i)? (f i ) () I req is chosen as the value for statistical evaluation. p.25.2.5..5 P(HO /I) III. POWER ONTROL In 2nd and 3rd generation mobile communication systems the transmitter power is controlled to minimize the co-channelinterference within the same system to ensure the quality of each link. Because of these controlling processes the interference power caused at a receiver in a different system varies depending on the interference situation in the interfering system. The power control (P) model used in the Monte arlo Simulator for MFS evaluation follows the algorithm specified in the GSM5.8 standard [3]. The transmitter power is controlled depending on the received signal strength and on the interference level at the corresponding receiving station. The principle of the P algorithm is depicted in Fig. 2. The interferences p(i ) self /I in Power ontrol min min < /I < < < out Fig. 2. The power control algorithm I adjacent /I max I self caused by transmitters of the same system are modeled by a distribution function. We found from simulations that for the examined Systems (GSM, TETRA) a normal distribution matches best the /I distribution function within one System. In the case of the perturbed link, the interferences I adjacent of the interfering system in an adjacent frequency band are also taken into account. Thus, the carrier-to-interference ratio I Rx at the perturbed receiver is I Rx = max in I self + I adjacent ; (2) where in denotes the received carrier signal strength before power control. The P algorithm compares the I Rx with threshold values for the minimum required and the maximum allowed =I level. If the =I value exceeds (falls short of) the given threshold, the 5 2 25 3 35 4 /I [db] Fig. 3. Quality handover decision probability transmitter power of the corresponding station is decreased (increased) until the received signal power meets its thresholds and as long as the received =I is out of the specified range. IV. HANDOVER To find a handover-model that is suitable for the Monte arlo method we examined the behaviour of the GSM5.8 handover algorithm proposed in [3]. The behaviour of the handover procedure is described by the probability of a handover action conditioned by the victim link =I. As such probability cannot be determined analytically it has to be obtained by simulations. We used the event driven GSM Simulation Tool GOOSE [4] to find the probability for a handover depending on the carrierto-interference ratio at the receiver. GOOSE provides realistic propagation models and simulates the GSM5.8 handover algorithm and its appendant protocols within a cellular network. onsidering the mobility of the mobile stations the carrier-tointerference ratio is calculated for each mobile station. Thus, the =I ratio can be measured for each handover action. One yields the probability P (=IjHO) for the =I conditioned by an handover action. Applying the Bayes Theorem the probability P (HOj=I) for a handover conditioned by the =I is P (HOj=I) = P (=IjHO)P (HO) : (3) P (=I) Because of the scaling on the =I distribution the handover probability is nearly independent on the factors that determine the =I, the morphology of the scenario, respectively. Figure 3 shows a measured probability for a handover action in the GSM system assuming an urban scenario with 37 hexagonal cells. For a carrier-to-interference ratio below 6 db no handovers could be observed because either the handover had taken place at an higher =I level or the call was cancelled because of a too high bit error ratio. In the M simulation the victim receiver changes the frequency with this handover probability depending on the calculated =I value. To model the calls where no handover can be initiated because of a too high bit error ratio it is assumed that the channel isn t changed for some stations although the =I ratio is too low. In this case this low =I is taken for evaluations. The frequency change is assumed to enlarge the distance to the minimum frequency separation. As shown in Fig. 4 the victim
... 2 2 3 4 5... >= khz f allocated allowed change channel if handover required guard band forbidden Fig. 4. hannel management swap channel if no free channel available system can be assigned groups of channels to take into account the frequency planning process for minimizing the co-system interferences. The TETRA system performs a cell-reselect protocol that can interrupt the call for up to 3 ms. The TETRA standard does not specify a quality-based cell-reselection decision algorithm. The TETRA algorithm is based on signal strength and threshold values that determine various decision parameters. As shown below, the ability to perform a handover because of too high interferences is an important requirement when using potential MFS bands for carrying traffic. A. MFS between GSM and TETRA This section presents and discusses some simulation results where the systems GSM and TETRA are contemplated. To show the impact of handover and power control on the inter-system interference, the TETRA cell-reselection procedure is assumed to behave like the GSM5.8 algorithm. For this, the measured curve depicted in Figure 3 has been scaled to the required inband /I of 9 db as specified in the respective standard [5]. Table I depicts the general parameters used in the simulations. The curves presented below show the cumulative distribution function of the value determined by equation. The =I difference on the x-axis applies exactly at that moment at which the required =I req ratio is still achieved. Thus, the probability of an inadequate coverage corresponds to the probability value along the curve at =I difference=. To show the impact of power control and handover on the intersystem interference the interference caused by a GSM mobile station at a TETRA hand-held is evaluated. The border frequency f b between the GSM uplink and the TETRA downlink band is 95 MHz. Thus, the carrier frequency of the first TETRA downlink channel is DF T ET RA fc = f b + MF S + W T ET RA =2 (4) whereas W T ET RA is 25kHz and denotes the bandwidth of one TETRA channel. The channels are numbered beginning at zero for the first channel adjacent to the MFS. The numbers increase with increasing distance to the MFS. Since only the channel adjacent to the MFS (No. ) is used this section presents simulation results for a worst case scenario. B. power control Fig. 5 shows the impact of the power control on the intersystem interference. Assuming an MFS of 6 khz between the GSM uplink and the TETRA downlink band the probability of a link failure is about 9%. onsidering for both systems a power control algorithm as described in section III the link failure probability decreases to about 2.3%. It should be noted.. no protocols victim and interferer power control. - -5 5 5 2 25 3 /I-difference [db] Fig. 5. Impact of power control on the interference General simulation parameters number of considered interferers 4 (nearest) density of 2/km 2 (GSM), interferers 2/km 2 (TETRA) coverage radius 2m (GSM), 2km (TETRA) Tx Power (max) 33 dbm (MS), 45 dbm (BS GSM) 49 dbm (BS TETRA) Tx Power (min) 3 dbm sensitivity -2 dbm (MS GSM), level -4 dbm (BS GSM), -6 dbm (BS TETRA), -3 dbm (MS TETRA) antenna gain 4 db (BS), -2 db (MS) antenna heigh 3m (BS),.5m (MS) TABLE I SIMULATION PARAMETERS that the interferences in Fig. 5 come from a microcellular system. Therefore, the transmission power of the interfering mobile station can often be lowered by the power control algorithm. Macrocellular interfering systems originate more interferences because of a higher mean transmission power to overcome the higher mean pathloss.. handover In Fig. 6 the =I-difference for the same scenario with an MFS of 6 khz and no protocols is compared with the difference assuming no MFS but handover. If a handover action is required because of a too low =I, the perturbed system changes to channel, that is,5*25khz apart from the outer frequency of the interfering system. It can be seen that even with no MFS the probability of a link failure can be decreased by applying handover with a suitable frequency planning. If the interferences come from a microcellular system with power control the outage probability decreases to less than.5%. Nevertheless, with the decreasing outage probability the number of handovers increases.
DF no protocols. handover (), no guard band handover, power control, no guard band.. - -5 5 5 /I-difference [db] 2 25 3 is :43 :39 P (HOjchannel) = (6) =6 One can conclude that a guard band of *25kHz is necessary to avoid needless handovers because the first channels can nearly never be used. For the other channels one computes the values given in Table IV-D. P (HOjchannel) represents the channel P (channeljho) P (HOjchannel) -9 4.3% -8 2.8%.647 9-64 %.23 65-.5%.6 D. apacity Gain Fig. 6. Impact of handover on the interference P(channel HO).. To quantify the capacity gain that can be achieved by using potential MFS bands for carrying traffic, the usability of each channel has to be determined. Therefore, the probability P (channeljho) is measured by statistical evaluation of the channel the perturbed station is transmitting on if a handover (HO) is performed because of too high interferences. For a MFS of khz this probabilityis shown in Fig. 7 for the scenario with and without power control. The probability P (channel) that a mobile station is assigned a channel follows an equal distribution. In Fig. 7 6 channels are taken into account 2, thus P (channel) equals =6. The probability that a channel is unno power control victim and interferer P TABLE II PROBABILITY THAT A HANDOVER IS REQUIRED BEAUSE OF INTER-SYSTEM INTERFERENES probability that a channel cannot be used because of too high inter-system interference. Thus, the capacity gain that can be achieved by using those channels for carrying traffic can be calculated by (N ) = N? X (? P (HOjchannel = i)) (7) N i= (N ) denotes the percentage of the MFS that is not perturbed if the MFS has a bandwidth of N channels. For the values P (channeljho) depicted in Fig. 7 the capacity (N ) is shown in Fig. 4. Up to 3 percent of the MFS with 2 channels can be used to carry traffic. As the first ten channels can nearly never.9.8.7.6 victim and interferer P (N).5.4 no power control. 2 4 6 8 channel (No.) Fig. 7. Density function of the channels conditioned by the handover action usable because of too high interferences is P (HOjchannel) = P (channeljho)p (HO) P (channel) whereas the values P (channeljho) can be taken from Fig. 7. It is obvious that the number of handovers decreases with increasing distance to the border frequency. The cascade shape of the curve is caused by the definition of the transmitter and receiver masks. For the shown situation P (HO) is 39% without and 27% with power control. It should be noted that we regard a worst case scenario. If we analyze the curve without power control the probability that one of the first ten channels is unusable 2 channels..4, 43, 47, 5, 53... 3 (5).3.2. 2 4 6 8 Fig. 8. Density function of the channels conditioned by the handover action be used, () approximately equals zero. Therefore, a MFS of less than channels width leads to a higher handover rate but no capacity gain. Thus, the channels should be reserved. It should be noted that the full capacity can only be reached if all required handover can be performed, i.e. if no handover is blocked. Especially, the handover blocking rate can be high if one cell only has a small number of frequencies. In this case one has to swap channels of the perturbed and an unperturbed station with a longer distance to the next interfering station and thus lower interference power. N
A. Spectrum Masks V. MFS BETWEEN UMTS AND DET As the specification of the air interface for the terrestrial part of UMTS is not finalized yet, the values defining the transmitter mask for an UMTS transmitter cannot be taken from standard specifications. The Figures 9 and represent the spectral power density of a TD-DMA base station and mobile station transmitter, respectively. These masks are based on values given in [6] and [7]. They have been derived from models of real wideband amplifiers, including their non-linear properties. required /I [db] 2 - -2-3 -4-5 -6-7 - -5 frequency [Mhz] 5 Power Density [dbc/728khz] Power Density [dbc/728 khz] -2-4 -6-8 - -2-4 TD-DMA BS -6-4 -3-2 - 2 3 4 frequency [MHz] - -2-3 -4-5 -6-7 -8-9 -4 Fig. 9. Transmitter Mask for a TD-DMA Basestation -3-2 TD-DMA MS - 2 3 4 frequency [MHz] Fig.. Transmitter Mask for a TD-DMA Mobilestation Fig.. Receiver Mask for a DET Station represents the ratio between the received signal power and the interference power measured in a 728 khz bandwidth, respectively. For instance, 5 MHz apart from the carrier the interference power measured in a bandwidth of 728 khz shall be at least -4 db below the DET carrier power. The minimum required inband /I is db. B. MFS alculation This sections presents the results for the MFS calculations between the UMTS and DET bands as depicted in Fig. 2. DET MFS 88 MHz 9 MHz UMTS Fig. 2. Allocation of Frequency Bands The DET system is located in the band 88 MHz to 9 MHz, the UMTS in the band 9 MHz and beyond that. The MFS is located in the spectrum foreseen for the UMTS. Since there are no receiver masks for TD-DMA available only the impact of the UMTS on the DET system is investigated although both systems interfere with each other because of their TDD transmission. The main simulation parameters are listed in Tab. V-B. The maximum distance between a DET transmitter and receiver is m whereas the cell radius of the UMTS is chosen to be one kilometer. It should be noted that the sim- The values for the spectral power density in Fig. 9 and Fig. represent the power measured in one DET carrier bandwidth. For example, the power transmitted by a UMTS mobile station in an spectrum interval of 728 khz width with the center frequency of the measurement bandwidth 2 MHz apart from the UMTS TX carrier frequency is -48 dbc. The required performances of DET in presence of an interferer are determined in [8]. The /I receiver mask considers the values given in the standard for interference performance, blocking, outband spurious emissions and inband spurious emissions. The maximum allowed interference power due to one of these issues are given for a specified signal power. These values define equivalent /I ratios. The maximum required /I values are taken to construct the mask depicted in Fig.. This mask Parameter DET (BS/MS) UMTS (BS/MS) cell radius m m antenna height m/2m 3m/.5m antenna gain 2dB 4dB carrier spacing 728 khz 6 khz TX power 24 dbm 4dBm/3dBm TABLE III SIMULATION PARAMETERS ulation results for the interfering mobile stations are valid for a worst-case scenario with interfering UMTS stations per km 2. Furthermore, no power control is taken into account. The four interfering stations nearest to the perturbed DET station
are considered. For an outdoor scenario Fig. 3 shows the distribution function of the difference between the required /I ratio and the /I for different MFS values. F(/I-difference).. MS, khz BS, khz MS, 2 khz MS, 728 khz BS, 2 khz BS, 728 khz. - -5 5 5 2 25 3 /I-difference [db] UMTS MS/BS -> DET Fig. 3. UMTS perturbes DET: DF of /I-difference If no MFS is introduced, UMTS BS as well as UMTS MS cause unacceptable interferences with a probability of 7% to 8%. In the second DET channel this interference probability decreases to 5% and 2%, respectively. Even for a MFS of 2 MHz the probability of interference is above % for a UMTS MS perturbing the DET band. The DET system applies a dynamic channel allocation, i. e. any channel that provides an acceptable interference level can be used. The probability that the /I difference is less or equal to zero can be interpreted as the probability that a channel is unusable because of too strong interferences. Thus, the capacity loss caused by inter-system interference can be derived from the interference probability. As long as there remains a certain probability that a DET channel near the UMTS band can be used, the introduction of a MFS seems questionable. Referring to the the transmitter masks (see Fig. 9 and ) the UMTS mobile stations cause more interference than the base stations. Fig. 4 shows the interference probability measured in the whole DET band ( channels), i.e. the capacity loss caused by the UMTS uplink. For No MFS this capacity loss caused by the interferences in the first two DET channels (see Fig. 3) is about %. If the interferer density decreases to 5 active stations per km 2 the capacity loss is about 8% and for 2 stations per km 2 this value is about 7%. From Fig. 4 one can also esti- F(/I-difference). 2 /km. - -5 5 5 2 25 3 /I-difference [db] 2 5/km 2/km 2 UMTS MS -> DET Fig. 4. UMTS MS perturbes DET: DF of /I-difference with UMTS MS density as parameter mate the influence of UMTS power control on the inter-systeminterference. If the mean transmit power of the UMTS mobiles is lower due to power control, the interference probability is represented by the probability that the /I-difference equals or falls below a negative value, e. g. -5dB. It can be seen that in this case about % capacity can be gained applying power control in the interfering system with a mean transmitter power of 5dB below the nominal power. VI. ONLUSION In this paper a method has been presented that is suitable to assess the impact of inter-system interferences of mobile communication systems in adjacent frequency bands. Simulations showed that power control and handover protocols allow the systems to bypass critical interference situation and decrease the probability of link failure. The impact of the protocols has been exemplarily analyzed for the systems GSM and TETRA. Furthermore, it has been shown that the ability to perform a qualitybased handover is essential for utilizing potential MFS bands to carry traffic. In this case, there is no need of an MFS because of the intelligence of the systems. For special scenarios the MFS can be used to reduce the rate of handovers caused by intersystem interference. The capacity gain, that can be achieved due to the utilization of the MFS has been evaluated. In principle, no MFS is needed for systems performing an interference-adaptive dynamic channel allocation. For the DET system the capacity loss caused by the introduction of UMTS (TDD-mode) in an adjacent band has been analyzed exemplarily for a microcellular outdoor scenario. The loss of capacity is mainly caused by emissions from UMTS mobiles. As a consequence of these results no MFS should be planned for 2nd and 3rd generation systems as the rare source frequency can be used more efficiently by using this frequency band. The drawback of this consequence is the required intelligence of the systems that they can guarantee the wanted grade of service (GoS). A further step towards efficient usage of the frequency spectrum is the sharing of frequency bands that allows different systems to use the same frequency at the same location at different times [9], []. REFERENES [] M. Lott and M. Scheibenbogen, alculation of minimum frequency separation for mobile communication systems, in Proceedings of the EPM 97 and 3.Fachtagung Mobile Kommunikation, (Bonn, Germany), 997. [2] M. Lott, M. Scheibenbogen, and P. Seidenberg, alculation of minimum frequency separation for mobile communication systems, in Papers presented at the 2nd ost259 Meeting, no. OST 259 TD(97)46, (Lisbon, Portugal), 997. [3] ETSI/T GSM, Recommendation GSM 5.8, radio subsystem link control, March 99. [4] M. Junius, Leistungsbewertung intelligenter Handover-Verfahren für zellulare Mobilfunksysteme. Dissertation, RWTH Aachen, Aachen, Oktober 995. [5] ETSI, RES TETRA, prets 3 392-2/393-2, part2: Air interface, August 995. [6] D.. G. ETSI SMG24, Guard band analysis for td-cdma, December 997. [7] ER, Sharing and compatibility of umts with adjacent services. dect (88-9mhz), August 997. [8] ETSI, Radio equipment and systems (res); digital european cordless telecommunications (dect), common interface, part2: Physical layer, October 992. [9] Motorola Radio Research Laboratory, An etiquette for sharing multimedia radio channels. ontribution to ETSI EP BRAN, May 997. [] F, In the matter of amendment of the commission s rules to provide for operation of unlicensed nii devices in the 5 ghz frequency range. Report and Order, January 997.