Packet service in UMTS: delay- throughput performance of the downlink shared channel Flaminio Borgonovo, Antonio Capone, Matteo Cesana, Luigi Fratta

2 1. Introduction In the last ten years: IP applications has pushed the data traffic to grow quickly 2G cellular systems have heavily changed the way in which users access the network. The challenge of third generation mobile communication systems is to provide access for a wide range of multimedia applications and services.

3 1. Introduction UMTS: 3G mobile communication system developed by ETSI, extend the present GSM service to include multimedia. UMTS provides great flexibility and a variety of different physical and logical channel types. Several user rates and protections are possible by choosing suitable parameters. More implementation complexity.

4 2. UTRA basics (1/3) Two access scheme for the radio interface: W-CDMA scheme 60MHz for downlink and 60MHz for uplink 3,84 Mchips/s, 5MHz for each channel, QPSK modulation TD-CDMA scheme 35MHz for downlink and uplink Physical channels are defined by the associated spreading and scrambling codes. Spreading sequence : XORXOR input data XOR spreading code XOR scrambling code Spreading code of CDMA: channelization code Scrambling code of CDMA: PN code

5 2. UTRA basics (2/3) Transport channels: Dedicated channels (DCH) Devoted to the connection between a single mobile station and the UTRA Network. Mapped into two physical channel :DPDCH, DPCCH Common transport channel Broadcast channel (BCH), paging channel (PCH) Random access channel (RACH), forward access channel (FACH): control information or packet Common packet channel (CPCH), downlink shared channel (DSCH) : packet only

6 2. UTRA basics (3/3) Power control DCH: TPC (Transmit power control) symbols in each slot carry a command for increasing or decreasing DSCH: computed on the basis of the power of the DCH

7 3. Downlink packet data services (1/3) Three channels for downlink direction DCH: assigned to single users through set-up and tear down procedures, subject to closed loop power control and service such as voice. DSCH No set-up, tear down procedure Doesn’t carry power control signaling, but must have an associated active DCH FACH Shared by many users to transmit short bursts of data No DCH must be activated

8 3. Downlink packet data services (2/3) For real-time circuit traffic—DCH Well known results show that CDMA with closed- loop power control is very effective.[14][14] Efficiency can be further enhanced by using powerful FEC codes. [15][15] For packet service—DSCH Due to the burstiness, the number of interfering channels, its power level, errors can be more efficiently obviated by ARQ than FEC[16][17][16][17]

9 3. Downlink packet data services (3/3) it is interesting to investigate whether UMTS achieves the highest data throughput with circuit or packet switching technology For circuit switching, the additional of any further channel beyond the capacity cannot be accepted since the BER will increase For packet switching, occasional increases in BER over its target value can be to tolerated Because of the use of ARQ techniques This paper provide a quantitative evaluation of the performance of the different alternatives.

10 4. Simulator description (1/4) Propagation model The received power P r is given by The path loss L is expressed as Each cell is assigned a signal tree of orthogonal variable spreading factors, so that channels in the same cell are always orthogonal

11 4. Simulator description (2/4) Traffic model The performance of the UMTS downlink heavily depends on the input traffic characteristics. Users become active according to a Poisson point process of intensity λ

12 4. Simulator description (3/4) Receiver model Carrier to interference ratio: Block error rate: Assumes an ideal ARQ procedure When system operates far from capacity:  Increase power  Increase power on the same channel to maintain SIR  Retransmitting  Retransmitting the packets. maximum interference tolerable is attained with a channel traffic G less than 1, and a further increase in retransmission would causes a strong decrease in throughput

13 4. Simulator description (4/4) Power control model Closed loop power control mechanism Inner loop: controls the transmitted power to maintain the SIR at target value Outer loop: controls the SIR to provide a target BLER →provide different qualities to different services. Each channel cannot exceed a transmitted power of 30 dBm, whereas the overall power transmitted by a BS is limited to 43 dBm. SIR Proportionally reduced

14 5. Simulation results (1/4) Effect of codes-1 Channel codes : Convolutional codes The encoded bits depend not only on the current k input data bits but also on past input bits Coding rate: Decoding strategy for convolutional codes: based on Viterbi algorithm

15 5. Simulation results (1/4) Effect of codes-1 Light codes Heavier code

16 5. Simulation results (1/4) Effect of codes-2

17 5. Simulation results (1/4) Effect of codes-3 Wrongly designed: Require very low interference Reach 1, but throughput is limited G: 0.955~0.97

18 5. Simulation results (2/4) Effect of user traffic on downlink shared channel If the amount of information is lower than the space available, the efficiency is reduced due to the unused space. Consider sources that generate an average number of packets N p in the range from 1 to 25…

19 5. Simulation results (2/4) Effect of user traffic on downlink shared channel

20 5. Simulation results (2/4) Effect of user traffic on downlink shared channel

21 5. Simulation results (3/4) Effect of control channels Too many users would reduce the system throughput. Limit the number of DCHs. (User arrived system is queued and wait for DCH availability) Power control commands indicate changes in the transmitted power level Both the power P DCH and P DSCH must change in the same way. The SIR achieved after despreading on the two channels are related as:

22 5. Simulation results (3/4) Effect of control channels Interference generated by the related DSCH: (loss-of-orthogonality factor = 0.4 [20])[20] Assume

23 5. Simulation results (3/4)

24 5. Simulation results (3/4) Effect of control channels Trade off: interference and multiplexing effect SF, coding rate, =>maximum user number minimum delay

25 5. Simulation results (4/4) Effect of power control The closed-loop power control: introduced to increase the system capacity The burstiness of data transmission may jeopardize the gain achieved. DCH introduce additional interference Thus DSCH is compared with FACH.

26 5. Simulation results (4/4) Effect of power control

27 6. Conclusions (1/2) Present some preliminary results obtained by simulation on the delay-throughput curves Focus on the system parameters and channel configurations Closed-loop power control mechanism SIR increases as the speed of the physical channel increases It has been verified the mechanism is very efficient even with low interference protection If multiple physical channels is allowed Intra-cell interference is improved by the closed-loop power control mechanism

28 6. Conclusions (2/3) When the system operates with the highest bearable interference A backoff mechanism is crucial to let the system operate close to capacity. The use of DCH for power control with DSCH may lead to instability if the number of DCH is not limited. If low speed users are served A great reduction of capacity may be observed because of the minimum unit that a single user may use The reduced frame filling degree does not reduce the interference, but yielding a net decrease in throughput

29 6. Conclusions (3/3) In spite of the several limitations of packet switching, due to its intrinsic flexibility, better adapts to interference limited systems than circuit switching.

30 Reference [14]. Erlang Capacity of a power controlled CDMA system (1993) [15]. Throughput analysis for Code Division Multiple Access of the Spread Spectrum Channel (1984) [16]. Channels with block interference [17]. Retransmissions versus FEC plus interleaving for real-time applications: a comparison between CDMA and MCD-TDMA cellular systems (1999) [20].3 rd Generation Partnership Project, RF system scenarios, 3G TR , Dic. 1999

31 Normalized energy per information: