The Fundamentals of Cellular Radio System Planning

Introduction

In a cellular system, a large geographic area is divided up into cells. In each of the cells there is located a base station (BS), which is often, but not always, near the centre of the cell. The base station, via microwave links, etc., is connected to a central control centre for mobile validation, etc., and then to the public switched network. When a mobile radio, or more strictly called; a mobile station (MS), is within a cell, and is making or receiving a call, it communicates with the base station of that cell. When the mobile moves into another cell, it communicates with the base station of the new cell. There are two main components in mobile radio systems. The first is the radio interface, which allows users to wander while communicating via radio from a mobile station (MS) to a base station (BS), and the second component is a fixed network that links to the public switched telephone network (PSTN) or the integrated services digital network (ISDN). What makes public cellular radio complex is the control structure that enables the network to know where a MS is currently located, and to track it irrespective of whether the MS is making a call or not, with the proviso that the mobile station is switched on. The control mechanism is made possible by a set of protocols. These protocols enable MSs to register on the network, facilitate call set-up and clear-down, switch MSs between base stations (BSs) as they travel, control the radiated power levels, provide security (in some systems), and perform a myriad of other vital functions.

The number of users a network can support is fundamentally dependent on the common air interface (CAI) over which users communicate. User capacity is dependent on many factors, but the main ones are: the amount of spectrum the regulators allocate, the size of the radio coverage area from a BS, and the amount of interference a particular radio link can tolerate. On purchasing new equipment, the first primary concerns are how and where to site the BSs and how to manage their use with the radio spectrum which has been allocated and how to optimize the teletraffic for the equipment deployed.

Each BS transceives with a number of mobiles residing within its radio coverage area. This area is known as a cell. For the purposes of design, the cell is hexagonally shaped, and is a separate entity from its surrounding cells. In practice, BSs are deployed so that each cell partially overlaps with other cells in the vicinities of their boundaries. It is this overlapping of cells that permits the mobile station to maintain near continuous communications with the called party whilst moving between cells. Suppose a mobile commences a call a position S within say, cell C. During the call, the mobile is moving from cell C to cell D. At some point within the overlapping regions of the cells C and D, which is also called the handover (handoff) HO, the received signal level at the mobile will be below a system threshold and lower than the received signal level from the BS of cell C. Signalling which occurs in this overlapping region, between the mobile, the BS and the control centre, results in an instruction that communicates with the mobile to switch to a different frequency and have its communications handled by the base station in cell D. This change in frequency occurs before the mobile reaches the point in the overlapping cells that would cause communications to be broken.

Cells are arranged in clusters, and usually each cluster uses the entire allocated spectrum. In the simple case, the clusters are designed in a mosaic fashion so that the limited spectrum is repeatedly used over large geographic areas, with each cluster supporting the same numbers of users. In the more complicated cases, the cell sizes are different and the cluster sizes, that is the number of cells per cluster, are different. However, this discussion and this applet will consider only the simple case where the cell sizes are the same area and the cluster sizes contain the same number of cells. If, for example, the number of cells in a cluster is four, then the entire allocated bandwidth will be used for this cluster. This means that each of the four cells can be allocated a quarter of the available bandwidth. If each cell is labelled 1, 2, 3, and 4, then the cells in adjacent clusters will also have their cells labelled 1, 2, 3, and 4, where each numbered cell contains the same band of frequencies as its equivalent cell number in an adjacent cluster. Should cluster A, say have its cell 1, next to cell 1 in cluster B, then the problem of co-channel interference arises. This co-channel interference is contained to acceptable limits by keeping the distance at a maximum between cells containing the same bands. One of the objects in the planning of cellular systems is to maximize the distance between cells in different clusters, which operate on the same band of frequencies. Perhaps, now it can be seen why as a mobile travels from one cell to another, which may be in a different cluster, it is assigned automatically, a different channel (as each channel of the mobile operates on a different frequency).

When a mobile system starts to mature, and more and more MSs use the system, it does not take long to realize that each mobile starts to have difficulty in setting up a call (that is a call attempt is blocked). The administrators of the system are faced with the problem of increasing the capacity of the system to reduce the blocking probability. It is important to note that as a cluster uses all of the available channels, then if the cell sizes are decreased so will the cluster size decrease and therefore there will be an increase in the number of channels per unit area. The most effective way of increasing network capacity is to decrease the cell size, although the complexity of the network infrastructure increases. The capacity of a network also depends on the number of cells per cluster, and the fewer the cells per cluster, the greater the capacity. This is because with fewer cells more bandwidth can be made available at each BS, and therefore more channels can be deployed at each BS. The teletraffic which is measured in Erlangs, (one Erlang can be thought of as one telephone conversation that lasts for three minutes) , is non-linearly related to the number of channels at a BS, and a disproportionate increase in teletraffic is obtainable for a given increase in the number of BS channels. However, reducing the cluster size, that is reducing the number of cells per cluster, decreases the minimum separation between cells operating with the same band of frequencies, and therefore increases the amount of co-channel interference. Keeping the number of cells per cluster constant and reducing the cell size and therefore the reducing the amount of transmitter power each MS and BS requires for a set receive carrier-to-interference ratio, does not reduce the minimum separation between cells operating with the same band of frequencies. This applet, demonstrates all of the above-mentioned concepts, by permitting the cells per cluster to be changed, the size of the cells, and the number of channels allocated to the cluster. The number of channels allocated to the cluster is the same as the allocated bandwidth, as for example, if each MS channel is 30 kHz wide, then for a cluster bandwidth of 25MHz, there will be 833 channels available (actually in practice, 832).

Microcell design

The minimum separation (D_s) required between two nearby co-channel cells is based on specifying a tolerable co-channel interference, which is measured by a required carrier-to-interference ratio (C/I)_s. The (C/I)_s ratio also is a function of the minimum acceptable voice quality of the system. In an AMPS system, (C/I)_s is equal to about 18 dB (the point at which 75% of the users call the system "good" or "excellent") and the minimum required separation, based on (C/I)_s = 18dB, is about 4.6R, where R is the radius of the cell (to the point where two faces of the hexagon join). In a cellular system, the number of cells K in a cell re-use pattern is a function of the co-channel separation D_s. For D_s = 4.6R, then K = 7. This means that a cluster of seven cells can share the entire allocated spectrum. In each of the two bands allocated for cellular in the USA (824-849 MHz mobile/869-894 MHz base) with the mobile channel spacing of 30 kHz, there are 832 voice channels, which gives about 119 channels, on average, per cell. In 1991, the conventional cellular systems in use since 1984 began to reach their capacity in the larger markets. In order to increase system capacity, approaches based on the co-channel interference reduction factor (CIRF) q_s, were taken. The CIRF is defined for an AMPS system of cluster size K = 3 or 7, as,

where D_s is the minimum required distance between any two co-channel cells in a cellular system corresponding to the required carrier-to-interference ratio (C/I) received at both the cell site and the mobile unit in a cell. R is the cell radius, and K is the number of cells in a cell re-use pattern, or the number of cells in a cluster. K is also called the cell re-use factor. The above equation is derived from an idealized hexagonal cell layout and is commonly used. The applet permits verification of the above equation for K = 3 or 7, by increasing the length of a radial from a central cell, to reach into other clusters. The length of this radial in units of R when extended to reach a cell in another cluster with the same number (same cell band) as the central cell gives a measure of the minimum required distance between any two co-channel cells. It is easily seen that for a cluster size of seven cells, where there are say seven clusters, a distance of 4.6R must be traversed before another cell of the same number is reached. The applet also shows that for a given number of channels per cluster, that the number of channels per cell m, is given by,

Calculation of Carrier-to-Interference ratio

The

in an AMPS system is based on two requirements. The first is, that all co-channel cells are at a distance of 4.6R away from the serving cell, and second is that, the value of q_s = 4.6 . This second requirement is based on a C/I of 18dB, where the interference is received from six co-channel cells in the first tier. To calculate the C/I the following equation is used;

for j = 1,2,…k

where R₁, is the radius of the serving cell, D_j is the distance to the next cell at tier j, with the same number (frequency band) expressed in terms of R₁, and n is the number of cells at this radius. At a different distance from the serving cell, another tier (tier 2) will contain cells with the same channels, and again at another tier, etc., until tier j is reached.

For example, if a seven cell cluster is taken on the applet, then there are six cells (i = 6), surrounding the serving cell at a distance D₁= 4.6 R₁ . If there are no further tiers (clusters outside of these seven surrounding the central cluster), then,

If further tiers exist, it is expected that this would reduce the 18.73 dB to the required 18 dB used in the AMPS system. The applet permits, for different cluster sizes, the values of D_j to be determined for different tiers, and thus, the value of C/I for a given configuration.

The Applet

First, the scroll-bars must be initialized. That is, before placing anything on the white canvas, values must be entered into the scroll-bars. There are five active scrollbars, and one that is not used. The first scrollbar on the top-left hand corner is initially the most important, as it determines the size of the cell. The cell radius R, is equal to any side of the hexagon. For convenience, its units are pixels on the screen. To get a reasonable configuration that would fit on the screen, the pixel size initially should be chosen around 15. The second top scrollbar is used to set the radius of the circle from the serving cell. Its units are given in terms of the cell radius R. The third scrollbar sets the number of clusters in the configuration. The minimum value is one, and the maximum is seven. The first of the lower scrollbars sets the number of cells per cluster, that is, it is the cell re-usage factor K. It ranges from one to seven. The second scrollbar in the lower set is reserved for future modifications of this applet, and is not used. The third scrollbar of the lower set allows the number of channel for the cluster to be entered. This represents, in terms of channels, the entire bandwidth allocated to the system.

The printouts on the screen, show the value of m, the number of channels per cell, and the unit area (the area of the cell size in square units of 100 pixels per side). More will be said about this printout in a moment. The bottom right-hand of the canvas printout, shows the mobile channels/cell/unit area.

Once the cell size has been set, along with the values required in the other scrollbars, clicking the mouse anywhere in canvas area will set the centre of the serving cell at the point where the mouse was clicked. By changing the cell radius, up or down, and reclicking on the canvas, the new configuration will appear. It is advisable to reclick the mouse on changing any of the active scrollbars, to obtain a picture of what the new configuration looks like. If the number of cells per cluster is chosen as 3 or 7 (K=3, or K = 7), a prompt will appear on the left-hand side of the canvas showing that the minimum separation distance is 3R or 4.6R, according to the first of the above equations. As a new cluster is added to a configuration, its colour in most cases will change. The colours used for individual clusters are; black for the first cluster, blue, red, green for the others.

Using the applet

To demonstrate change in capacity as cell size changes

First, the cell radius must be increased to 62 pixels, showing the Unit Area = 1, (actually 0.999671). This sets the proper normalization. If the actual area was 10 square km, then all calculations would be scaled to this value, by first normalizing to unity. After this, the total channels per cluster is set. The value of channels per cell is read as being true, and the mobile channels/cell/unit area will provide the traffic density. By reducing the cell radius, the configuration of the clusters and cells per cluster, as well as the total channels per cluster will not change, but, the mobile channels/cell/unit area will increase. This, demonstrates how reducing the cell size, the traffic density increases. This effect can be seen for different configurations, such as cells per cluster, channels per cluster, etc.

To determine minimum separation distance between the serving cell and the co-channel cells (cells with the same number and consequently the same bandwidth allocation).

The use of the second top scrollbar permits a circle of varying radius to extend to the centre of any cell on the screen. By extending this circle to the centre of a co-channel cell, the radius of this cell can be read-off from the scrollbar value. This allow C/I calculations to be carried out according to the equation given above. It can be observed that for a seven sector, seven cell cluster, that the radius is 4.6R as indicated by the printout on the left-hand side of the canvas, when this configuration is chosen.

The applet has been designed to be as flexible as possible within the constraints of seven cells per cluster, and seven clusters. Hopefully, something will be learnt from it.

The source code (version Rev.1 98/08/08) is available according to the GNU Public License.

Tony Townsend, tonyart@ieee.org

Return to Applets page.