Information spreading on third generation Mobile Communication System
IOANNIS KOSTOPOULOS
Information and Communications Department
Siemens S.A
Athens, GREECE
CONSTANTINOS GIANNARIS
Department of Information Systems
London School of Economics
London, U.K
Abstract: - The purpose of this paper is to present the coordination of the information transmission in a UMTS cell between base stations and mobile stations. Uplink and downlink directions must be coordinated between a mobile station and the base station. Additionally, the transmission between different mobile stations of a cell and the base station must be coordinated.
Keywords: Interference, signalling, capacity, frequency efficiency, channel power, path loss, cell.
1 Introduction
For spreading subscriber information, a unique code is provided for each subscriber. This code is referred to as the spreading code. The linkage of the high bit rate code with the original subscriber information transforms the original signal into a broadband signal. This broadband signal is transmitted together with broadband signals from other subscribers using the same frequency band over the radio interface. The receiver receives the sum of all these signals. By relinking the summation signal with the (synchronized) subscriber code the original subscriber information is regenerated. The remaining information stays in its broadband form and therefore constitutes an underlying signal.
2 Main System Features
Power Control:
Power control is an important means in all radio communication systems to master the interference situation. To keep the interference level low, the signal power of each user should not exceed substantially the value necessary for good quality.
In a CDMA system power control is a vital measure to keep the system running. It is probably the most important feature in the uplink. Imperfect power control directly affects the uplink capacity. Without fast and accurate power control the capacity of a CDMA mobile radio system would be degraded significantly. As each user disturbs the links of all other users in the base station receiver, the power control must try to let no user's signal incoming at the base station get stronger than that of any other user. The received level of the farthest (and thus weakest) user in the cell will determine the actual signal level of all users. The consequence is that in large cells the information level becomes lower with respect to the thermal noise level. This fact has crucial influence on the portable payload and on the size of a cell.
Interference:
Signal quality is in general determined by the C/I-ratio or, respectively in digital communication systems by the ratio Q between the energy of a received information bit Eb and the composite power of thermal noise power spectral density No and total interference spectral density It.
Q = Eb / (No + It) (1)
The spreading process in spread spectrum systems makes interference power also noise like, thus we can summarize:
No + It = Nt and Q = Eb / Nt (2)
The tolerable minimum quality Q depends on the modulation method, the radio channel characteristics (delay spread), the information bit rate and, in spread spectrum systems, the spreading gain, which is the ratio of the spreading code chip rate W and the information bit rate Rb. In the UMTS system the required Q-ratios are different for up- and downlink and depend on the various voice and data services and scenarios.
In CDMA systems each user is disturbed by all other active users of the whole system, not only by the co-cell users because the same carrier frequency is allocated to all cells. The Q-ratio decreases with growing cell size, as by power control measures the received signal level (Eb) of all users is adapted to the received level of the weakest, i.e. the farthest user. Power control of course reduces the interference power (It-part of Nt) from all co-users as well, but the receiver thermal noise power (No-part of Nt) remains constant and gradually becomes dominant with growing cell radius, and from a certain size on no user interference additional to thermal noise is tolerated. Q would drop below the allowed minimum. The cell capacity is then, even in an isolated cell, equal to Zero.
In very small cells thermal noise may be neglected, as it will be surpassed by far by the weakest received signal level. Cell capacity reaches a maximum value, the so-called asymptotic capacity N. Not regarding the effects of power control, frequency reuse and link activity the maximum uplink capacity would be:
N = 1 + (W / Rb) / (Eb / Nt) (3)
Frequency reuse and imperfect power control result in a reduction of this capacity value.
Frequency Efficiency fe:
In a CDMA network all cells use the same carrier frequency. A user will be disturbed not only by the co-users of the same cell but also by the users of the surrounding cells. Interference coming from the users of the own cell is called intracell interference; the one coming from all other cells is called intercell interference. The cell capacity in a network is substantially reduced with respect to the capacity of an isolated single cell. The capacity degradation caused by frequency reuse is called frequency efficiency.
Hand-Off HO:
Each user in a CDMA network may have not only one (main) server base station (best server), but, if he is close to the border of neighbour cells, will be served by two or three or even more stations. This feature supports the so-called soft handover function and it improves the signal quality in the downlink because it helps to overcome the negative fading influence. In the link budget this is taken into account as HO-gain. But on the other hand soft handover significantly reduces the network capacity, as users in HO-status contribute to the load of several cells.
Signalling Overhead:
The main signalling channels in a CDMA network are the pilot-, sync- and paging channels.
The percentage of power used for signalling purposes reduces directly the maximum payload. Signalling should not require more than 10...15% of the available power; even 7...8% may be sufficient under favourable conditions.
3 Spread Spectrum Techniques
CDMA systems are based on "spread spectrum multiple access" (SSMA) techniques. This is a technique where a carrier which is already modulated by user information is modulated a second time in that way as to generate a spread wide band signal. The signal obtained after spreading has a very low power spectral density, but contains the same energy in a larger bandwidth. At the receiver this power density may be comparable to the noise level. The inverse operation is followed in the demodulator to despread the received signal and recover its original shape.
In CDMA systems or "direct sequence" (DS) SSMA systems the user information bits are modulated with a PN (pseudonoise) sequence of much higher bit rate than the original information signal. Each digital unit of this spread information is called a chip. The chip rate determines the occupied bandwidth. Multiplying the spread signal at the receiver with the same PN-sequence will reproduce the original narrow band information with high power spectral density. All received wide band signals which had not been spread with the applied despreading code remain spread with low power spectral density and are noise like. The wanted signal can be distinguished out of a large number of other signals. The spreading gain (or system gain) G is proportional to the ratio between chip duration Tc and information bit duration Tb.
G (dB) = 10 * log (Tb / Tc) =
10*log (W / Rb) (4)
A further very important effect is the inherent antijamming capability of this method. The despreading process at the receiver does not only separate the wanted signal from all other user signals, but it also spreads the spectrum of narrow band interferers, thus reducing their spectral density.
4 Interference and Cell Load in a Cellular Network
The carrier frequency of each cell is reused in all other cells. This causes a reduction of the uplink capacity by a factor fe. The total interference level increases by a factor f (= frequency reuse factor) because of interference from other cells:
Iother-cells = Isame-cell * f (5)
The total interference Itot is the sum of Isame-cell + Iother-cells;
Itot = Isame-cell + Iother-cells = Isame-cell * (1+f) (6)
The total interference density will thus be
Itot = Itsame-cell *(1 + f ) (7)
The factor f depends on the propagation slope, the log-normal standard deviation and on the multiple hand-off status of the cell.
The frequency efficiency fe is defined as
fe = 1 / (1+f) (8)
In common multicell environment with 3 available HO-cells the resulting reduced uplink capacity is about 60% of the capacity in an isolated cell, i.e. the frequency efficiency fe = 0.6.
There is a close relationship between power control and "soft handoff" in DS-SSMA systems. Outside of an inner circle around a base station where this station dominates all others, each mobile user must be served by several stations. Soft handover is based on maintaining communication to at least two base stations at the same time. The same information will be sent to and received from both stations. This principle has severe consequences on the manageable cell load, but in the context with power control improves considerably the handover process:
When a mobile has communications to two base stations 1 and 2, and the connection to BS1 would decay whereas the connection to BS2 would remain stable, it would receive different power control orders from both of them. BS1 would indicate the mobile station to raise its power, while BS2 request to lower it after the received power had received a certain level. The algorithm in the mobile station raises its power only when both stations are calling for it. While it will lower the power when just one of them notifies it. In our example the quality of connection to BS1 would get worse and finally would be abandoned. The mobile station would remain connected to BS2 until the next handoff.
To start a second communication for getting in soft handover a mobile station must not raise its power too much before having a second connection. Simulations showed that on average a mobile station is in soft handoff for about 40% of time. This means that a big part of the system resources is engaged in soft handoff traffic that affects of course the system capacity.
5 Differences between Uplink and Downlink
Eb/Nt-ratio:
The downlink may be characterised as a one-to-many connection. All channels are synchronized by the base station. They may be considered to be coherent.
The uplink is a many-to-one connection. All users transmit unsynchronised. Their signals must be considered non-coherent.
To maintain a certain link quality the bit error rate BER must not be inferior to a minimum value. The decisive parameter for BER is the Eb/Nt-ratio. In a Rayleigh fading environment, due to their diverging degree of coherence, the tolerable Eb/Nt-ratios for the same received quality are different in up- and downlink.
Ec/Nt-ratio:
In a well balanced network the pilot range (= downlink range) shall slightly exceed the uplink range. But low output power of the base station may cause the downlink to be the limiting factor. The total available BS power has to be shared by all downlink traffic channels and control channels (pilot + sync + paging channels). The pilot channel is the main consumer (~ 5 - 10% of the total power). The total required signaling overhead may go up to about 15% of the downlink power. The remaining power is left for traffic. The pilot quality is defined by the ratio Ec/Nt.
Ec = Eb*Rb/W (9)
6 Relations between Cell Load and Cell Size
The following equations reflect the fundamental relations between the planning parameters and the network features.
Definition of the notations used in the formulae:
Wchip rate
Rbinformation bit rate
Ebenergy of one information bit
Ecenergy of one chip in the spread (wide band) signal
Nothermal noise spectral density
Iofraction of the DL interference spectral density originating from the own cell
Iocfraction of the DL interference spectral density originating from other cells
Ittotal interference spectral density
Nttotal interference plus thermal noise spectral density
Pc power control factor in the UL (0....1; should be >0.8)
vlink activity factor (~0.5 for speech)
frfrequency reuse factor (= 1/fe –1 with fe = frequency efficiency)
HOhandoff reduction factor; (value >1) reduces directly the capacity; if 30% of the users are in the handoff status, the HO factor is approximately 2.
ηfraction of available BS-power used for signalling (pilot + sync + paging)
δfraction of available BS-power used for pilot
NDLtheoretical maximum number of users in the DL
Nasymptotic capacity (theoretical maximum channel number in the UL)
Mnumber of traffic channels per cell; x = M/N = cell load
Tr(inverse) transmission loss as a function of cell load or traffic channels per cell
PL(inverse) path loss=
Tr * Gant_rx *Gant_tx
PmobMobile transmission power
Ptcelltotal available BS transmission power
Uplink asymptotic capacity N:
There are fixed relations between the capacity and the size of a cell. The so-called asymptotic capacity (or pole capacity) of the uplink defines the theoretical maximum number of users. It neglects thermal noise, i.e. the signal power of each user received in the base station exceeds by far the thermal noise level; this is a cell with Zero size.
(10)
Downlink asymptotic capacity NDL:
Thermal noise and interference from other cells are neglected in this formula.
(11)
Uplink transmission loss TrUL as a function of user channels per cell M:
The permitted signal attenuation is limited by the number of users; with only one user the limit would be set by the thermal noise.
(12)
Downlink transmission loss TrDL for traffic channels as a function of user channels per cell M:
A limiting barrier for the permitted DL attenuation is reached with a rather low number of traffic channels because of the signalling overhead and the interfering influence from other cells.
(13)
Transmission loss for the pilot channel Trpilot as a function of user channels per cell M:
For the determination of link balance the uplink range for traffic channels is compared to the downlink range of the pilot channel.
(14)
Received channel power RxUL in the uplink:
The uplink-received level of each user traffic channels corresponds to the incoming power of the weakest (farthest) user. It is a function of the user load but not of the antenna gain and of the Mobile power, because these figures modify the allowed cell size (corresponding to TrUL) in a way to keep the received signal power at the minimum level.
(15)
7 Results and Recommendations
The PL graph shows the relation between load and maximum path loss for UL-traffic channels, DL-traffic channels and Pilot channel. The curves for the traffic channels disclose decreasing path loss if the number M of channels increases.
Furthermore the traffic channel curves show a limiting barrier (pole with Zero
path loss) as soon as the denominator of the applied transmission loss formula vanishes ( = Zero). This threshold indicates the theoretical maximum channel load of the link. It may happen that in the displayed diagram the threshold is suppressed due to the granulation of the scale on the M-axis. The threshold for the path loss of the UL channels indicates the 100% load situation, when M = N = asymptotic capacity.
The barrier for the path loss of the DL traffic appears at a relatively low number of channels because of the interference from other cells (Ioc) and the power spent for the signalling channels. The tolerable load could be raised by reduction of the required Eb/Nt ratio, but this would end with a lower signal quality in the DL.
No breathing is recognizable for the path loss of the pilot channel. This link is independent of the channel load and only limited by the power reserved for the pilot. The pilot range is the decisive DL criterion for link balance. In link balance considerations the UL traffic range is compared to the pilot range in the DL.
The user should not be surprised if e.g. halving of the power control factor Pc does not entail a path loss reduction of 3dB in the UL. Pc-modification affects the asymptotic capacity of the uplink N as well, which on its part appears also in the formula of the uplink transmission loss.
It could be surprising that the UL channel received power does not change if, with unchanged user load, a mobile changes its output power or if the antenna gain is modified. The explanation is that these variations directly modify the cell size correspondingly and the UL Rx power is adapted to the modified signal range.
8 Conclusions
As the demand of mobile services increases, the number of channels assigned to a cell eventually become insufficient to support the required number of users.
In this paper, we described the main features of a CDMA system, the spread spectrum multiple access technique and the antijamming capability.
We analysed the interference, the load of a cell in cellular network, the differences between uplink and downlink and the relation between the cell load and the cell size.
The results produced from the simulations show that the size of the cell follows the fluctuating cell load.
Reference:
[1] V.K.G. Garg, K.F. Smolik, J.E. Wilkes, Applications of CDMA in Wireless/Personal communications, Feher / Prentice Hall digital and wireless communications series (1997)
[2] A.J. Viterbi: „CDMA: Principles of Spread Spectrum for third Generation Mobile Communication (1995),
[3] T. Ojanperä, R. Prasad: Wideband CDMA for third Generation Mobile Communication, (1998)
[4] R. Prasad, W. Mohr, W. Konhäuser, Third Generation Mobile Communications Systems, Artech House Publishers (04/2000)
[5] G. Calhoun, Third Generation Wireless Communications, Artech House Publishers (07/2000)