WO2001072071A1

WO2001072071A1 - Method of defining network cells in a communications network

Info

Publication number: WO2001072071A1
Application number: PCT/RU2000/000095
Authority: WO
Inventors: Evgeny Petrovich Vishnevsky; Alexandr Sergeevich Anikonov; Vladimir Grigorievich Maslov; Konstantin Jurievich Ushakov; Stanislav Evgenievich Vishnevsky
Original assignee: Motorola Inc.
Priority date: 2000-03-21
Filing date: 2000-03-21
Publication date: 2001-09-27

Abstract

The present invention relates to a method of communications network planning comprising the steps of: forming (8) a plurality of cells (6) depending on grade-of-service requirements for each cell, allocating (9) a base station (1) in each of said cells (6), assigning (10) traffic channels to each of said cells (6), assigning (12) frequencies to each of said cells (6). The step of forming (8) a plurality of cells (6) comprises the sub-steps of: determining (13) a power domain (2, D(p))(6), which includes all points for which the received power (P¿r?) is greater than a predetermined threshold power (Pr,th), determining (14) an snr domain (3, D?(snr)¿)(6), which includes all points for which the signal-to-noise ratio (snr) is greater than a predetermined threshold signal-to-noise ratio (snr¿th?), determining (15) a Shannon domain (4, D?(sh)¿)(6), which includes all points for which the channel capacity (C) is greater than a predetermined signaling transmission rate (R¿8?), determining (16) a BER domain (5D?(BER)¿)(6), which includes all points for which the error probability (P¿e?) is less than a predetermined error probability (Pe,th), creating (17) the cell (6) as an area of the intersection of the power domain (2, D?(p)¿), snr domain (3, D(snr)) Shannon domain (4, D(sh)), and BER domain (5, D(BER)).

Description

Method of defining network cells in a communications netw ork

Field of the invention

The present invention generally relates to the method of planning a cellular network, and in particular though not exclusively it relates to defining network cells in a high population urban area

Background of the Invention

The method of cellular network planning is described in the patent application which is assigned to the assignee of the present invention and which is hereby incorporated for reference. The central problem of cellular network planning is the effective cell formation (CF), especially in a high population urban environment under pre-specified grade-of-service (GOS) constraints. The GOS is mainly determined by the quality and reliability of the communications in the network and cellular network planning (CNP) involves consideration of threshold values of the received signal power P_{r lh} and signal-to-noise ratio snr_lh (power GOS), error probability P_{e lh} (information GOS). Key factors in high population urban areas (where traffic demand is concentrated) are propagation conditions of the electromagnetic waves (EMW) and distribution of communications channels in time

In other words, the formation of network cells is essentially influenced by the EMW propagation conditions and by the expected traffic demand and occupation of channels of the network. In prior art methods of forming network cells the propagation conditions and the channel occupation are allowed for by simple statistical and stochastic models. Conventional cell formation procedures are based on various cell definitions. E.g. in WO 90/10342 "Method for planning radio cells" by Gunmar K. and in WO 93/15591 : "Method and Apparatus for planning a cellular radio network by creating a model on a digital map adding properties and optimising parameters based on statistical simulation results ' by Markus O the cell is detined as an area within which the field strength of the base station exceeds a predetei mined threshold value The requirement of an acceptable signal-to-noise ratio is disclosed e g by Gunmar K in US 5 307 10, "Method for determining the degree of co\ erage in the mobile radio sy stem" In most of the works the classical definition of cell is used when the radio regions are designed in accordance to the distributions of the radio traffic densitλ in the area of interest e g in "The cellular concept Bell S\ st Tech J 58, ( 1 ) pp 1 5-43, 1 979 by MacDonald V H and in L S 4 667 202 Mobile radio network bv Kammerlander K et al and in US 5 465 390 "Method for laving out the lnfiasti ucture of cellular communications network' bv C ohen R and in US 4 759 051 "Communications system' b\ f lan K and in US 5 428 817 Mobile communications sv ste having variable cov erage areas ' b\ Yahagi M

Flowever these simple models reflect the problems in realization of the network inadequately only and thus fail to provide optimum solutions in particular as to the formation of network cells

Especially the conventional cell definitions do not allow for the reliability GOS constraint so that the conventional cell formation process is not optimal As a result the subsequent BER control in conventional cellular networks often requires an increased transmitted power or the repetition of the cell formation process with other system parameters (I e an iterative procedure) Further, the use of statistical models of EMW propagation and regular cell pattern of the area of interest (AOI) with the simplest cell shapes (triangular, rectangular, hexagonal) leads to overestimation of power, frequency and technical resources in conventional CF approaches due to an imperfect accounting for the above key factors that constraint the CF in high population urban areas The low resolution of conventional DTM used in some approaches (about ten meters or higher) and the appropriate CF methods do not permit a high accuracy in high population urban environment Interactive iterative procedures, used in some conventional CF approaches, become cumbersome and unrealistic for micro-cell network planning in large high population urban areas when the number of micro-cells is big enough (hundreds or more) The problems of inadequate wave propagation models and channel stochastic become even more prohibitive in a network with micro-cells of a decreased cell size e g of about 200m, which operate at low power of about 20m W w ith antennas located at lampposts etc 5

The present invention seeks to enable a more accurate cell formation in such networks in order to mitigate or avoid the disadvantages and limitations of the prior art

10 As a first aspect of the inv ention a cell formation method according to claim 1 -■ provided

Another aspect of the invention is a communications network s\ stem with mιcκ - cells in a high population urban environment as defined in claim 1 1

: 5

The dependent claims are directed to preferred embodiments of the invention

Preferred embodiments of the invention will now be described bv way of example only, with reference to the accompanying drawings 0

Brief Description of the Drawings

Fig 1 shows an example for the domains of which a cell according to this invention is formed 5

Fig 2 shows an explanatory diagram for the determination of contributions by electromagnetic wave reflections

Fig 3 shows an explanatory diagram for the determination of contributions by 0 electromagnetic wave multiple reflections Fig 4 shows an explanatory diagram for the determination of the effect of shadowing of electromagnetic wav es

Fig. 5 shows an explanatory diagram for diffraction effects of electromagnetic waves

l' ig 6 shows an explanatory diagram for the determination of the basic contributions to a received signal by electromagnetic wav es which are diffracted on side edges of a building

Fig. 7 shows an explanatory diagram for the quantitative determination of reflection effects of electromagnetic waves

Fig. 8 shows schematically the flowchart of an embodiment of the cellular network planning method according to the present invention.

Fig. 9 shows the step of forming a cell according to the present invention in detail.

Detailed Description of a Preferred Embodiment

The method of communications network planning according to the present invention comprises the steps of forming a plurality of cells depending on grade-of- service requirements for each cell, allocating a plurality of base stations in the AOl. assigning traffic channels to each of said cells, assigning frequencies to each of said cells, wherein the step of forming a plurality of cells comprises the steps of determining a power domain ( D^{{ p)} ) for each cell, which includes all points for which the received power ( P_r ) is greater than a predetermined threshold power ( P_{r lh} ), determining an snr domain ( Dⁱ ""^r) ) for each cell, which includes all points for which the signal-to-noise ratio ( .snr ) is greater than a predetermined threshold signal-to- noise ratio ( snr_lh ). determining a Shannon domain ( D^{{ >l>} ) for each cell, which includes all points for which the channel capacity ( C) is greater than a predetermined signaling transmission rate ( R_s ), determining a BER domain ( D^u"^'''^{< ]} ) for each cell, which includes all points for which the error probability ( P. ) is less than a predetermined error probability ( P_L, _lh ). and creating the cell as an area of the combination of all of the power domain, snr domain. Shannon domain, and BER domain.

The communications network system according to the invention comprises a plurality of cells, wherein the size of each of the plurality of cells is determined by all of a power domain ( D^{{ ,, )} ). a snr domain ( D ^wr> ). a Shannon domain ( D ^h) ) and a BER domain ( D ^BER) ).

The CF problem is of particular practical interest for optimal micro-cellular radio network planning in an urban environment. Due to its efficiency the power, frequency and technical resources may be saved in the course of CNP. handoff operations are facilitated, and performance studies of the entire planned cellular radio network are enabled by highly accurate simulations which are based on a refined digital terrain model (DTM) which will briefly be discussed below.

In Fig. 1 an example of a cell 6 is shown which is constituted by the intersection of various basic domains, each being characterized by a specific parameter. In Fig. 1 a base station 1 is established at a location within the envisaged cell 6. The signal from this base station 1 will be received with a quality that varies with the location of the receiver (not shown). The quality of the received signal is expressed as "grade-of- service" (GOS), and areas with a satisfying GOS with respect to a specific parameter are shown in Fig. 1 . E.g. the domain with a satisfying GOS as to the power of the received signal is designated by 2 and will be referred to in the following as power domain D^ip) ; the domain with a satisfying GOS as to the signal-to-noise ration of the received signal is designated by 3 and will be referred to in the following as snr domain Z)⁽ "^r) ; the domain with a satisfy ing GOS as to the channel capacity as opposed to the demand of traffic is designated by 4 and will be referred to in the following as Shannon domain D^{( p)} , and the domain with a satisfying GOS as to the bit error rate (BER) of the received signal is designated by 5 and will be refeπed to in the following as BER domain D^{{ p)}

According to the invention in a communications network system w ith a plur ality of cells 6 as shown in Fig 1 each cell is deteimined by the above domains 2. 0 4. and 5 Thus, in general the size of each of the plurality of cells 6 is determined b> all of a power domain 2 ( D^{{ p)} ), a snr domain 3 ( D "^r) ), a Shannon domain 4 ( Z^{)( /,)} ) and a BER domain 5 ( D^(B" )

In particular each of the plurality of cells 6 is formed as the intersection of said domains, such that each cell 6 corresponds to the intersection area of all of said power domain 2, D^(p) , said snr domain 3, D^, '"^r) , said Shannon domain 4. D⁽"^h) and said BER domain 5, D^(I>,R)

In the following the method according to the invention shall be described by which the determination of the domains and eventually of the cell is achieved, with reference to Fig 2 to 9

A flowchart for a software based on a method of communications network planning according to the present invention is shown in Fig 8 The method comprises the steps of reading all collected data on the specific network demands and the geographrc particularities of the network area T hese data are the basis for the planning procedure and are compiled in step 7 of the flowchart In step 8 a plurality of cells 6 is formed depending on grade-of-service requirements for each of said cells 6 After the cell has been formed, in the general procedure in step 9 base statrons 1 are allocated in the AOl, and in the following steps 10 and 12 traffic channels are assigned to each of said cells 6 and frequencies are assigned to each of said cells 6, respectively. In order to allow for long term changes of traffic demand in step 1 1 a query is performed whether or not new traffic data are avarlable If so, the procedure branches back to the status immediately before the channel assignment in step 10 and the channel assignment is repeated so as to optimize the network efficiency in the whole Else the frequency assignment in step 12 is carried out

However, the above method is a rather rough approach and the formation of the cells in step 8 remains the central problem of network planning If therefore the planning is based on data with a rather poor resolution in the prior art the resulting network will be all but optimized

According to the invention it is thus suggested to perform the following steps when tυiming said plurahtv of cells 6 These sub-steps are shown in Fig 9 The flowchart of Fig 9 is entered at step 13 (from step 7 in Fig 8) and is continued (after step 17) by step 9 in Fig 8 First, a brief overview of Fig 9 will be giv en, the way of how to determine each of the domains will be described below

For forming a cell 6 in a first sub-step 13 a power domain 2 D^ip] is determined The power domain D⁽ⁿ⁾ includes all points of an area, for which the received power P_r is greater than a predetermined threshold power P_{r lh}

In the next step 14 of the procedure an snr domain 3 D^{ "^r) is determined The snr domain D ^mr) includes all points of an area, for which the signal-to-noise ratio snr is greater than a predetermined threshold signal-to-noise ratio snr_h

In step 1 5 a Shannon domain 4 D^{( h)} is determined, which includes all points of an area for which the channel capacity C is greater than a predetermined signaling transmission rate R

At last, in step 16 a BER domain 5 D^{{KI R)} is determined, which includes all points of an area for which the error probability P_c is less than a predetermined error probability P_{L lh} Fventually. in step 1 7 the domains 2 to 5 are intersected so as to form the desir ed cell 6 These above steps 13 through 1 7 of determining the domains and of intersecting them are repeated for all of said network cells 6 It w ill be understood that the above order of the sequence of steps 13 to 16 is not mandatory but can freely be chosen, being completed by step 1 7 in any case

In other words, in the present in ention the cell is defined as the intersection of power, snr . Shannon, and BER domains, each calculated for a giv en base station site in an area of interest. In the power domain the received signal pow er P_r ≥ P_{r lh} . m the snr domain the signal-to-noise ratio snr ≥ snr_lh . in the Shannon domain the signaling transmission r ate R < . C being the Shannon channel capacit . and in the BER domain the bit error rate P_ι ≤ P_{t lh} is calculated

The above threshold values are preset before starting the planning procedure on the basis of the expected and desired characteristics of the communications network Assessing and setting of threshold v alues is known to the one skilled in this art and will therefore not be explained here any further

This cell definition allows optimization of the cell formation process and allows for the above GOS constraints just at the beginning of the cellular network planning. The exact calculation of the domains relies on the use of a high resolution model of the network geography To that order a refined digital terrain model (DTM) of high resolution is used to provide the required high accuracy of cell formation and to account for key factors that determine cell formation in high population urban areas With such a DTM a realistic radio channel model is used for the received radio signal represented as the sum of the deterministic and statistical parts. The former usually includes a few principal ray-theoretical components that mainly contribute to the total received signal power (about 90% or more), and may be predicted with high accuracy in each point of AOl via the DTM of high resolution by exact accounting for the main physical mechanisms of EMW propagation. The latter is the sum of signals of low power which forms the conventional Rayleigh fading signal This channel model may be considered as an extension of conventional Rice channel model to a few particular stable components in the l eceived signal

The mam idea of the threshold domain method is to obtain directly highlv accurate boundaries of the cell components by operating in a vector format of DTM thus avoiding the bulky point-to-point calculations for formation of the cells

In the following a detailed description of the determination of the domains will be given

The pow er domain 2 D ⁾ is one of the cell components that follows fr om the above extended cell definition This domain corresponds to the power GOS constraint P_{r lh} (a power threshold of the received signal) with a given base station 1 of the coordinates M_Q = (x₀ , y₀, z_Q ) The domain is defined as

D ^p) = {(x. y) e

, =)} (1 )

where P_r is the power of the signal received in a point M(x,y,z_r) In accordance with the DTM-based multipath dynamic channel model the received power may be expressed by

K being the extended Rice factor

In other words the received power P_r is regarded as a combination of a deterministic power ( P_{r ιkl} ) and a fading power ( P_{r fa l} ) and is described by the extended Rice-model

From ( 1 ) and (2) P_del may be assessed as

Thus, the power domains D^(p) in terms of the power of the deterministic (stable) part of the received signal are'

T he number N_ιkl of principal rays in the deterministic part of the received signal is determined by environmental particularities of the pair M.M_Q). Thus, each point M is characterised by the appropriate value N_Jel(M,M₀) In practice N_ιkl is in the order of one or ten, and it is assumed that in the whole area of interest

N_ma being a parameter of the desired cellular network.

From the (4) and (5) it follows that

D (/^») _. U (6)

D_m n D„ = 0, (m ≠ n)

where

It should be emphasized that in each domain D, (/ = 1,2,..., N_ma. ) one and only one 1-ray channel model is realized.

In real high population urban environment the received signal is of multi-ray nature mainly due to mirror reflection from walls of buildings. Thus, domains D, are considered as consisting of individual "reflection" ( R, ) - areas where the signal reflected from a certain reflecting surface could be received (with predetermined location of the base station, of all reflecting surfaces and obstacles). Any deterministic ray can be received only inside its reflection. Any reflecting surface in couple with real or imaginary source bears its reflection.

Let M be the total number of reflections R_t in the AOL Then the reflecting surface corresponding to reflection R, as a, and the source coordinate is denoted (x_0l, y_0l , z_0ι) . With the auxiliary areas G;'} _k the following relations hold:

where the number of indices , j, k etc. is equal to the number of rays received in every point of considered area which for simplicity is limited by 3. Domains D, are expressed as follows: A ⁼ U^G.⁽¹ T ( 1 )

where 7, " denotes an area where power received from the source located in the point (x_0; , y_Qι . z_0/ ) is not less than /^ in absence of obstacles;

Thus the received power is calculated as a combination of few reflection rays or as a combination of direct rays ( P₀ ) and reflection rays ( R_t |_{; ifl} ).

T_t ⁽ ' is the area where the over all power received from two sources located in points (x₀, , y_Qι , z_0ι ) and (x_0l , y_0l , z_Ql ) is not less than P_{r lh} .

T^^} is the area where the over all power received from three sources is not less than P_{r lh} . Algorithms for formation of reflections R and areas T are described below.

The reflections R are determined by the following expression:

where A_t is a sector of reflection for the z^'-th source, determined as shown in Fig. 2.

In Fig. 2 the sector of reflection is shown. The boundary of this sector is determined by the height h_l of reflecting wall a, . (All heights: ., . z_0l , etc. are measured from MT level). For A, > z_0l . we obtain /, = ∞ and a boundary of R_t coincides with the AOl boundary It is assumed that . = 0 corresponds to LOS area and R₀ coincides with the whole AOl

In case of multiple sequential reflections the sector R could be determined by repeated application of the abov e procedure The case of double reflections is shown in Fig 3 with the reflected sector R in case of a double reflection

Areas b, in ( 10) ai e areas under a /-th building S_A ' is the shadow as cast bv the 0th building, in which the /-th source is not to be receiv ed These aiea could be constructed by conventional means of computational geometrv Following the usual approach it is assumed that all buildings may be described bv l ectangular prisms with an arbitrary polygon in the base The subtraction of S_A in ( 1 0) should be implemented in two steps First shadows of buildings located inside 4 , are constructed, like for the building GFIIJ in Fig 4

But some buildings may be located between the source and reflecting surface casting a shadow after reflection as, for instance the shadow of the building CDEF in Fig 4 In this case it is necessary to "reflect" such a building in a reflection surface and to construct a shadow of "reflecting building as if it were a real one, using an imaginary source In Fig 4 C'D'E'F' is a perimeter of "reflected" building and

AA'"D"E"E'" is a part of its shadow inside the reflection from the imaginary source BS Thus, such "reflected" buildings are preferablv included in the logical intersection over k in (10)

Thus, in the above preferred embodiment the reflection rays (R) comprise also multiple reflections ( R, )

A further aspect of the determination of a power domain is diffraction It is considered as a correction effect to the shadow, with the effect of "decreasing" the building shadow First the diffraction caused by roofs of buildings will be considered with reference to Fig 5, in which the diffraction by building roofs is shown In Fig 5 S is a source, ABCD is a building Diffraction corrections are determined as follows The distance EF (diffraction shortening d of the shadow in Fig 5) is determined by the following system of equations'

(11)

Here F(V) is a Fresnel function for which the usual appioximation was determined, λ is the wavelength, and p is used as follows

p = ^P {4πP_th)-X (12)

where

P ¹ ιh = P ¹ r lh (13)

4π

In practice for small values of d the following approximation is valid

with an inverse Fresnel function approximated in the following way

„.,/ x 0.225 _{n <}

F a) = - , a <0.5 a (15)

Diffraction on side building edges are considered in a similar w a\ as shown in Fig 6 in which the correction of diffraction on side building edges is explained The same formulas ( 1 1 ) - ( 15) as above are used for calculation of shadow boundary shirt-, d in this case In case of an arbitrary n conditions for areas T^"' can be written a- follows

Σ ( 16)

(^ - 0, )' + ( - ><. ) ^* P

wheie / is a combination of all reflection factors corresponding to the z-th source is determined by ( 12) In case of two rays ( 16) is easily resolved for ;

This equation determines one or two bounded contours, which are approximated by polygons with rather a great number of edges In general the boundary of X"' is determined as will be explained in the following

The equation (16) is equi alent to an algebraic equation of degree 2n, which can be solved for x for any arbitrary value of y For a set of y_k with equal distance between each other all real solutions x,(y_k ) of ( 16) are obtained bv an appropriate numerical method After that all points of x, {y_k ) are linked to the closest points oi ø Uø₊i ) . if the difference between them is less then the difference between different x, {y_k ) Points which are left free (not linked) by now will now be linked with each other This operation leads to one or several bounded non-crossing contours These contours determine a solution of the problem If obtained contours are not bounded o^r are self-crossing r is increased and the above procedure is repeated

An example of determining the required domains D, is shown in Fig 7 The onh reflecting surface considered is a and its reflection R, is dashed in Fig 7 R corresponds to the whole AOl G^ coincides with R, T^^] is described by oval in

Fig 7 The contribution of G^^{( ]} in D₂ (doubly dashed) is determined by crossing of

G_Q^ and T_Q ²⁾ So, a practical implementation of the described procedure consists in elementary operations on areas (regions) and could be realized by conventional means of computational geometry

Thus, in the above preferred embodiment the received power is calculated as combination of direct rays, reflection rays and diffraction contributions

The determination of the power domain 2 in step 13 is, thus completed and in the following step 14 the snr domain 3 is determined The domain IX "^r) corresponds to the snr GOS constraints snι_lh For a given base station 1 of the co-ordinates M₀(x₀,}_Q z₀) the snr domain is defined as the area which satisfies

D "^r} = {(x,y)e AOl | sm (M,

(18)

By definition

nr=^- (19) P,

where P_n is the noise power which is assumed to be constant in the AOl From (18) - (19), for any point M e D^{s"^r) we get

P_r{M,M₀)=P„ snr(M,M₀)≥P_n mr_ώ (20)

By virtue of equation (20), the calculation of D^{"^r) is reduced to calculation of the appropriate power domain with the power threshold

snr_lh (21)

which is given by the refined DTM technique The determination of the snr domain 3 in step 14 is thus completed and in the following step 15 the Shannon domain 4 is determined. The Shannon domain Z ⁽ ' is defined as the one in which the basic Shannon requirement for reliable communications is satisfied. Precisely, for a given BS site M₀(x₀,y_Q.z₀)

D^[

AOI \ C(M,M₀ ) > R, . Λ/ = (.v. v. =_r )] (2 )

where 72, is the signaling transmission rate and C is the Shannon capacity of the radio channel for the pair (_V , ₀) .

The analytical expression of C for the DTM-based dynamic multipath channel model is given bv

C{M,M₀) (23)

where

Here P₀ is the transmitted power, a, = afM, M₀) is the path loss for the z^'-th principal ray and W is the bandwidth. It is to be noted that (23) is a generalisation of the Shannon formula for the multipath channel with stable components in the received signal.

To obtain the required Shannon domain £>^{( Λ)} without cumbersome point-to-point calculations the expression (23) is transformed into a function of SNR. Since

P + P snr = fad ι + κ p del n (25; . K P..

where

X, = ∑XP₀ (26)

it results from (24),(25), and ( 26) that

Rewriting the last term in (23) and using (27) equation (28) holds

To transform the first term in (23) the approximation is made

P = a.²P_n = P (29)

where

Then

Taking into account (28). (31 ), and (25) the required expression of C as a function of SNR is:

i.e. C = C_M (snr) .

It is the basic formula for calculation of the required Shannon domain D^{( h)} by the threshold domain technique.

As mentioned above, N_del = l,2,...,N_max in AOL So:

D^W = {jD, (33)

/=!

where D, is the power domain corresponding to the appropriate power threshold value

P_r „(l) = snηP„ (34)

Here snη is a solution of the equation ( _l {sm ) = R (35)

where the term on the left hand side is given by (32) with N_/c, = /

Thus, the calculation of Shannon domain D^{l h)} is reduced to calculation of powei domains D, with appropriate power thresholds P_{r lh} (l) and step 1 5 is completed

In this embodiment of the method of cellulai net oi k planning the last sub-step of determination of threshold domains consists in calculating the BER domain i e step 1 6

The lefined DTVI multipath channel model is used for deriving the lequired analytical expression for Shannon capacity and bit-error-rate, and for the high accurate calculations of cell components which will be explained in the following

A BER domain D^{U>I K)} is defined as the one where the pre-specified information GOS constraint, giv en by a probability error threshold P_{L lh} . is satisfied

D^(β,R) = {(x. y ) e AOI \ P_L {M. M₀) ≤ P _lh , M = (x, y, z_r)} (36)

P_L (M,M₀) being an average error probability for the pair (M. \/₀) Calculating the BER domain inv olves the following aspects

The refined DTM-based multipath channel model is used with in general a few deterministic components considered as stable part of the received signal The channel is treated as conventional Rice fading channel The above solutions are expressed in terms of SNR This permits to derive appropriate SNR-thresholds for pre-specified information GOS constraints and, consequently, reduce the problem to the determination of an snr domain, which was explained above After step 16 of d termining the BER domain step 8 is completed by step 1 7 namely creating the cell area e g by intersecting the so far determined domains An essential step of the cellular network planning procedure has thus been finished, and the cells 6 may be regarded as being of optimized shape and size T he degi ee of optimization strong depends on the efficiencv of the refined Digital l en ain Model (DTM) that is employ ed for determination of the abov e domains In the following the diffei ence between the l efined Digital Terrain Model according to the inv ention and the well known Digital Flevation Model (DEM ) for general surfaces such as atmosphei e lav ei s gi oundwater tables etc w ill be explained In a DEM a real surface is approximated bv a set of points in a three-dimensional C aitesian fi ame (X Y Z) where the X- and V -axis represent geographic cooi dinates ( I e longitude and latitude respectively ) and the Z-axis represents the altitude abov e sea lev el T hese (X Y 7 ) triplets are pi eferablv arranged in a grid based DEM with rows and columns Λ single point can be accessed bv its row and column number that coπesponds to piedefined geographic coordinates (l e latitude/longitude UTM-coordinates etc) Othei structures which may be used for the elevation data are Triangulated Irregulai Networks (TIN) and vector based structures

In the refined DTM according to the invention terrain specific information is derived from elevation data for the area to be inv estigated (with the data being presented as a DEM) so as to create a set of digital terrain specific information data including the elevation data itself

In the refined DTM each item in a map such as buildings, tiees and streets is represented by a square a polygon, etc and it is stored in a memory (The data are preferably stored in the memory as vector drawing data ) Additionally a specific height is added to the ground plan (square polygon) of the item and is also stored in said memory In a first preferred embodiment of the refined DTM the specific height is the average height of the building over its ground plan In a second preferred embodiment of the refined DTM the specific height of the building is its maximum height Other definitions of the specific height of a building will be obvious to persons skilled in the art So a "2D+" model results that is slightly reduced in comparison to a complete 3D model, in that a specific height is stored as an attribute to the respective ground plan of an object rather than a third co-ordinate Z varying over the ground plan and requiring additional memory and processor performance 27

A further simplification of the refined Digital I errain Model is the neglect of foliage The data in said DTM may thus be organized in three lay ei s

1 2D layei of the ground plan of buildings, each building hav ing a specific height assigned to it

2 2D lay er of foliage and trees (circles or other cuives) iepresent g the size and location of the ti ees and foliage

3 2D layer of road network

The coordinates ai e the same foi all layers and thus lav ei s are of the same size and correspond to each another

With the r efined DTM an exact and simple determination of the abov e power domain 2 ( D^{( l )} ). snr domain 3 ( D^{""^r) ). Shannon domain 4 ( D^l ) and BER domain 5 ( D^{U1, K)} ) becomes feasible, referring to a plurality of ground plans and specific heights of a plurality of buildings within said cell only (instead of a set of complete 3D data) Further, due to its efficiency the cell formation according to the present invention allows to overcome the main disadvantage of conventional cell formation, namely the overestimation of the required power, frequency, and technical resources for the cellular network because of an insufficient consideration of factors that limit cell formation in a high population urban environment

The proposed method of cell formation in high population urban areas improves the efficiency of utilisation of power, frequency and technical resources The improv ements as accomplished by this invention as opposed to kno n cell formation technologies may be summarized as follows

1 The method guarantees pre-specified GOS values (power, sni and BER) in each cell, thus, making the CF process optimal In particular, this is v ery important for call loss (CL) performance of mobile communications systems using the above cells In such a case, there is only one cause of CL no available traffic channel m a source cell for call blocking (CB) or in a target cell for call dropping (CD) In contrast, additional causes of CL arise in a real world situation where the conventional regular cells are currently used Precisely, the above power GOSs are not satisfied in some domains within a regular cell Extensive GSM-simulations have shown that these additional factors of CL can essentially degrade the CL performance of such systems This is a serious problem to ov ercome which arises in conventional regular cellular systems There is no such problem when a method of CF is used in CNP

2 The method accounts for the above key factors that constraint the CF in high population urban environment I e it takes into account the complex topological stiuctui e of dense areas on the basis of a refined high resolution DTM (resolution of 2m or less), it accounts for the fine structure of electromagnetic (EM) fields in a complex urban environment bv exact consideration of ma phy sical mechanisms of EMW piopagation (diffraction and multiple reflection), it relies on a realistic model of a multipath channel predictable with high accuracy in every point of the AOl for an arbitrary BS site

3 The cell formation is performed automatically, no interactive iterative planning procedures or field strength measurements are used for cell formation, interactiv e simulation tools are used only as auxiliary means for visual illustration of the results and their analysis

4 The DTM-based threshold domains technique of CF gives directly fnghlv accurate cell boundaries under the predetermined GOS values, thus, avoiding the cumbersome point-to-point calculations in prior art CF processes

While the invention has been described in terms of particular structures, devices and methods, those of skill in the art will understand based on the description herein that it is not limited merely to such examples and that the full scope of the invention is properly determined by the claims that follow Reference Numerals

base station power domain ( D^(p) ) snr domain (D'⁴"") Shannon domain ( ^(>/,) ) BER domain (D^l" ) cell = intersection of all domains read input data cell formation step base stations allocation channel assignment query: new traffic data available? frequency assignment determining a power domain (D^{p)) determining an snr domain ( ^{"⁾ ) determining a Shannon domain ( D^{( h)} ) determining a BER domain ( D^{BER) ) creating the cell of all of the power domain, snr domain. Shannon domain, and BER domain

Claims

. Method of communications network planning comprising the steps of: forming (8) a plurality of cells (6) depending on grade-of-service requirements for each cell, allocating (9) a base station ( 1 ) in each of said cells (6), assigning ( 10) traffic channels to each of said cells (6), assigning ( 12) frequencies to each of said cells (6). wherein the step of forming (8) a plurality of cells (6) comprises the sub- steps of:

determining ( 13) a power domain (2, D^(p) ) for each cell (6), which includes all points for which the received power PX is greater than a predetermined threshold power ( P_{r lh} ),

determining (14) an snr domain (3, D^{ "^r) ) for each cell (6), which includes all points for which the signal-to-noise ratio ( snr ) is greater than a predetermined threshold signal-to-noise ratio ( snr_lh ).

determining ( 15) a Shannon domain (4, D^{( h)} ) for each cell (6), which includes all points for which the channel capacity ( C ) is greater than a predetermined signaling transmission rate ( 72 ),

determining ( 16) a BER domain (5, D^{{I! R)} ) for each cell (6), which includes all points for which the error probability ( P_e ) is less than a predetermined error probability ( P_{e lh} ),

creating ( 17) the cell (6) as an area of the intersection of the power domain (2, D^(p) ), snr domain (3, D^{ "^r) ), Shannon domain (4, ^{( ,A)} ), and BER domain (5, D^{l,hR) ).

2 Method according to claim 1 , wherein the recerved pow er ( P_{r l} ) is a combination of a deterministic power ( P_{r lh du} ) and a fading power ( P_{r lh fad} )

3 Method according to claim 2. wherein the fading power ( P_{r lh /tl l} ) i approximated by the extended Rice-model

4 Method according to claim 2 or 3, wherein the received signal is calculated as a combination of direct rays ( 72, ) and reflection rays ( 72, |_;i0)

5 Method according to claim 2 or 3, wherein the received signal is calculated as combination of direct rays ( 72₀ ), reflection rays ( 72, |,_.₀ ) and diffraction contributions

6 Method according to any of claims 2 to 5, wherein the reflection rays (72) comprise multiple reflection rays ( P )

7 Method according to any of the preceding claims, wherein the channel capacity ( C ) is a function of said signal-to-norse ratio ( snr )

8 Method according to any of the preceding claims, wherein determining each of said power domain (2, D^{t p)} ), said snr domain (3, D⁽ '"^"' ), said Shannon domain (4, D^{sh) ), and said BER domain (5. D^{mR)) is based on a plurality of ground plans and specific heights of a plurality of buildings within said cell

9 Method according to claim 8, wherein each of said plurality of specific heights corresponds to the average height of each of said plurality of buildings

10. Method according to claim 8, wherein each of said plurality of specific heights corresponds to the maximum height of each of said plurality of buildings.

11. Communications network system with a plurality of cells (6). wherein the size of each of the plurality of cells (6) is determined by a power domain (2. D^(p)), a snr domain (3, D^{""^r)), a Shannon domain (4. D^{{ )}) and a BER domain (5, D^{(, R)}).

12. Communications network system with a plurality of cells (6) according to claim 11, wherein each of the plurality of cells (6) corresponds to the intersection area of said power domain (2, D^p)), said snr domain (3, D^{wr)), said Shannon domain (4, D^(h) ) and said BER domain (5, D^(BER) ).