Wcdma for umts radio access for third genergation mobile communacations phần 7 pps

identification time depends mainly on the number of cells and multipath components that the UE can receive, in the same way as with intra-frequency handovers. The requirement for the cell identification in 3GPP is 5 seconds with CPICH E c =I 0 > À20 dB [6]. 9.3.4 Summary of Handovers The WCDMA handover types are summarised in Table 9.9. The most typical WCDMA handover is intra-frequency handover that is needed due to the mobility of the UEs. The intra-frequency handover is controlled by those parameters shown in Figure 9.16. The intra- frequency handover reporting is typically event-triggered, and RNC commands the hand- overs according to the measurement reports. In the case of intra-frequency handover, the UE should be connected to the best Node B(s) to avoid the near–far problem, and RNC does not have any freedom in selecting the target cells. f1 f1 f1 f1 f1 f1 f1 f1 High capacity sites with two frequencies f1 and f2 f1 f1 f1 f1 f2 f2 f2 f2 f1 f1 f1 f1 f2 f2 Micro layer with frequency f2 f2 f2 f2 f2 f2 f2 f2 f2 f2 f2 Macro layer with frequency f1 f1 f2 f1 f2 Figure 9.31. Need for inter-frequency handovers between WCDMA carriers Measurement trigger is RAN vendor specific algorithm The more peaks the UE can receive with its matched filter, the longer the WCDMA cell identification takes. The cell identification time depends on • Number of multipaths • Number of cells within detection range • Number of already found cells • Size of the neighbourlist Cell identification typically <5 seconds. (1) RNC commands UE to start inter-frequency measurements with compressed mode. (2) UE finds P-SCH peaks. (3) UE identifies the cell with S-SCH and CPICH and reports measurements to RNC (4) RNC sends handover command to the UE Figure 9.32. Inter-frequency handover procedure Radio Resource Management 259 Inter-frequency and inter-system measurements are typically initiated only when there is a need to make inter-system and inter-frequency handovers. Inter-frequency handovers are needed to balance loading between WCDMA carriers and cel l layers, and to extend the coverage area if the other frequency does not have continuous coverage. Inter-system handovers to GSM are needed to extend the WCDMA coverage area, to balance load between systems and to direct services to the most suitable systems. An example handover scenario is shown in Figure 9.33. The UE is first connected to cell1 on frequency f1. When it moves, intra-frequency handover to cell2 is made. The cell2, however, happens to have a high load, and RNC commands load reason inter-frequency handover to cell5 on frequency f2. The UE remains on frequency f2 and continues to cell6. When it runs out of the coverage area of frequency f2, coverage reason inter-frequency handover is made to cell4 on frequency f1. Handovers are used in Cell_DCH state. In Cell_FACH, Cell_PCH and URA_PCH state the UE makes the cell reselections between frequencies and systems itself according to the idle mode parameters. The states are described in Section 7.8.2. Table 9.9. WCDMA handover types Handover type Handover measurements Typical handover measurement reporting from UE to RNC Typical handover reason WCDMA intra-frequency Measurements all the time with matched filter Event-triggered reporting – Normal mobility WCDMA ! GSM inter-system Measurements started only when needed, compressed mode used Periodic during compressed mode – Coverage – Load – Service WCDMA inter-frequency Measurements started only when needed, compressed mode used Periodic during compressed mode – Coverage – Load Intra-frequency Inter-frequency due to load Intra-frequency Mobile moves Inter-frequency due to coverage cell1 - f1 cell2 - f1 cell5 - f2 cell3 - f1 cell6 - f2 cell4 - f1 Figure 9.33. Example handover scenario 260 WCDMA for UMTS 9.4 Measurement of Air Interface Load If the radio resource management is based on the interference levels in the air interface, the air interface load needs to be measured. The estimation of the uplink load is presented in Section 9.4.1 and the estimation of the downlink load is in Section 9.4.2. 9.4.1 Uplink Load In this section two up link load measures are presented: load estimation based on wideband received power, and load estimation based on throughput. These are example approaches that could be used in WCDMA networks. 9.4.1.1 Load Estimation Based on Wideband Received Power The wideband received power level can be used in estimating the uplink load. The received power levels can be measured in the Node B. Based on these measurements, the uplink load factor can be obtained. The calculations are shown below. The received wideband interference power, I total , can be divided into the powers of own- cell (¼ intra-cell) users, I own , other-cell (¼ inter-cell) users, I oth , and background and receiver noise, P N : I total ¼ I own þ I oth þ P N ð9:3Þ The uplink noise rise is defined as the ratio of the total received power to the noise power: Noise rise ¼ I total P N ¼ 1 1 À  UL ð9:4Þ This equation can be rearranged to give the uplink load factor  UL :  UL ¼ 1 À P N I total ¼ Noise rise À 1 Noise rise ð9:5Þ where I total can be measured by the Node B and P N is known beforehand. The uplink load factor  UL is normally used as the uplink load indicator. For example, if the uplink load is said to be 60 % of the WCDMA pole capacity, this means that the load factor  UL ¼ 0:60. Load estimation based on the received power level is also presented in [8] and [9]. 9.4.1.2 Load Estimation Based on Throughput The uplink load factor  UL can be calculated as the sum of the load factors of the UEs that are connected to this Node B:  UL ¼ð1 þ iÞÁ X N j¼1 L j ¼ð1 þ iÞÁ X N j¼1 1 1 þ W ðE b =N 0 Þ j Á R j Á  j ð9:6Þ where N is the number of UEs in the own cell, W is the chip rate, L j is the load factor of the jth UE, R j is the bit rate of the jth UE, (E b =N 0 ) j is E b =N 0 of the jth UE,  j is the voice activity factor of the jth UE, and i is the other-to-own cell interference ratio. Radio Resource Management 261 Note that Equation (9.6) is the same as the load factor calculation in radio networ k dimensioning in Section 8.2.2. In dimensioning, the average number of UEs, N, of a cell needs to be estimated, and average values for E b =N 0 , i and  are used as input parameters. These values are typical for that environment and can be based on the measurements and simulations. In load estimation the instantaneous measured values for E b =N 0 , i,  and the number of UEs N are used to estimate the instantaneous air interface load. In throughput-based load estimation, interference from other cells is not directly included in the load but needs to be taken into account with the parameter i. Also, the part of own-cell interference that is not captured by the Rake receiver can be taken into account with the parameter i. If it is assumed that i ¼ 0, then only own-cell interference is taken into account. 9.4.1.3 Comparison of Uplink Load Estimation Methods Table 9.10 compares the above two load estimation methods . In the wideband power-based approach, interference from the adjacent cells is directly included in the load estimation because the measured wideband power includes all interference that is received in that carrier frequency by the Node B. If the loading in the adjacent cells is low, this can be seen in the wideband power-based load measurement, and a higher load can be allowed in this cell, i.e. soft capacity can be obtained. The importance of soft capacity was explained in radio network dimensioning in Section 8.2.3. The wideband power-based and throughput based load estimat ions are shown in Fig- ure 9.34. The different curves represent a different loading in the adjacent cells. The larger the value of i, the more interference from adja cent cells. The wideband power-based load estimation keeps the coverage within the planned limits and the delivered capacity depends on the loading in the adjacent cells (soft capacity). This approach effectively prevents cell breathing which would exceed the planned values. Table 9.10. Comparison of uplink load estimation methods Wideband received power Throughput Number of connections What to measure Wideband received power I total per cell Uplink E b =N 0 and bit rates R for each connection Number of connections What needs to be assumed or measured separately Thermal noise level (¼unloaded interfer- ence power) P N Other-to-own cell interference ratio, i Load caused by one connection Other-cell interference Included in measurement of wideband received power Assumed explicitly in i Assumed explicitly when choosing the maximum number of connections Soft capacity Yes, automatically Not directly, possible via RNC No Other interference sources (¼adjacent channel) Reduced capacity Reduced coverage Reduced coverage 262 WCDMA for UMTS The problem with wideband power-based load estimation is that the measured wideband power can include interference from adjacent frequencies. This could originate from another operator’s UE located very close to the Node B antenna. Therefore, the interference-bas ed method may overestimate the load of own carrier because of any external interference. The Node B receiver cannot separate the interference from the own carrier and from other carriers by the wideband power measurements. Throughput-based load estimation does not take interference from adjacent cells or adjacent carriers directly into account. If soft capacity is required, information about the adjacent cell loading can be obtained within RNC. The throughput-based RRM keeps the throughput of the cell at the planned level. If the loading in the adjacent cells is high, this affects the coverage area of the cell. The third load estimation method in Table 9.10, in the right-hand column, is based simply on the number of connections in the Node Bs. This approach can be used in second generation networks where all connections use fairly similar low bit rates and no high bit rate connections are possible. In third generation networks the mix of different bit rates, services and quality requirements prevents the use of this approach. It is unreasonable to assume that the load caused by one 2-Mbps UE is the same as that caused by one speech UE. 9.4.2 Downlink Load 9.4.2.1 Power-Based Load Estimation The downlink load of the cell can be determined by the total downlink transmission power, P total . The downlink load factor,  DL , can be defined to be the ratio of the current total transmission power divided by the maximum Node B transmission power P max :  DL ¼ P total P max ð9:7Þ Note that in this load estimation approach the total Node B transmission power P total does not give accurate information concerning how close to the downlink air interface pole capacity the system is operating. In a small cell the same P total corresponds to a higher air interface loading than in a large cell. Figure 9.34. Wideband power-based and throughput-based load estimations Radio Resource Management 263 9.4.2.2 Throughput-Based Load Estimation In the downlink, throughput-based load estimation can be effected by using the sum of the downlink allocated bit rates as the downlink load factor,  DL , as follows:  DL ¼ P N j¼1 R i R max ð9:8Þ where N is the number of downlink connections, including the common channels, R j is the bit rate of the jth UE, and R max is the maximum allowed throughput of the cell. It is also possible to weight the UE bit rates with E b =N 0 values as follows:  DL ¼ X N j¼1 R j Á  j ðE b =N 0 Þ j W Á½ð1 À " Þþ " ii ð9:9Þ where W is the chip rate, ðE b =N 0 Þ j is the E b =N 0 of the jth UE,  j is the voice activity factor of the jth UE, "  is the average orthogonality of the cell, and " ii is the average downlink othe r-to- own cell interference ratio of the cell. Note that Equation (9.9) is similar to the downlink radio network dimensioning (see Section 8.2.2.2). The average downlink orthogonality can be estimated by the Node B based on the multipath propagation in the uplink. The values of E b =N 0 need to be assumed based on the typical values for that environment. The average interference from other cells can be obtained in RNC based on the adjacent cell loading. 9.5 Admission Control 9.5.1 Admission Control Principle If the air interface loading is allowed to increase excessively, the coverage area of the cell is reduced below the planned values, and the quality of service of the existing connections cannot be guaranteed. Before admitting a new UE, admission control needs to check that the admittance will not sacrifice the planned coverage area or the quality of the existing connections. Admission control accepts or rejects a request to establish a radio access bearer in the radio access network. The admission control algorithm is executed when a bearer is set up or modified. The admission control functionality is located in RNC whe re the load information from several cells can be obtained. The admission control algorithm estimates the load increase that the establishment of the bearer would cause in the radio network. This has to be estimated separately for the uplink and downlink directions. The requesting bearer can be admitted only if both uplink and downlink admission control admit it, otherwise it is rejected because of the excessive interference that it would produce in the network. The limits for admission control are set by the radio network planning. Several admission control schemes have been suggested in [10–15 ]. In [10, 12 and 13] the use of the total power received by the Node B is supported as the primary uplink admission control decision criterion, relative to the noise level. The ratio between the total received wideband power and the noise level is often referred to as the noise rise. In [10] and [13] a downlink admission control algorithm based on the total downlink transmission power is presented. 264 WCDMA for UMTS 9.5.2 Wideband Power-based Admission Control Strategy In the interference-based admission control strategy the new UE is not admitted by the uplink admission control algorithm if the new resulting total interference level is higher than the threshold value: I total À old þ ÁI < I threshold ð9:10Þ The threshold value I threshold is the same as the maximum uplink noise rise and can be set by radio network planning. This noise rise must be included in the link budgets as the interference margin: see Section 8.2.1. Wideband power-based admission control is shown in Figure 9.35. The uplink admission control algorithm estimates the load increase by using either of the two methods presented below. The uplink power increase estimation methods take into account the uplink load curve (see, e.g., tsekkaa na ¨ ma ¨ ). Two different uplink power increase estimation methods are shown below. They can be used in the interference-based admission control strategy. The idea is to estimate the increase ÁI of the uplink received wideband interference power I total due to a new UE. The admission of the new UE and the power increase estimation are handled by the admission control functionality. The first proposed method (the derivative method) is presented in Equation (9.13) and the second (the integral method) in Equation (9.14). Both take into account the load curve and are based on the derivative of uplink interference with respect to the uplink load factor dI total d ð9:11Þ which can be calculated as follows Noise rise ¼ I total P N ¼ 1 1 À  ) I total ¼ P N 1 À  ) dI total d ¼ P N ð1 À Þ 2 ð9:12Þ ∆ L Interference level I threshold I total_old load η Estimated increase of interference ∆ I Max planned noise rise Figure 9.35. Uplink load curve and the estimation of the load increase due to a new UE Radio Resource Management 265 The change in uplink interference power can be obtained by Equation (9.13). This equation is based on the assumption that the power increase is the derivative of the old uplink interference power with respect to the uplink load factor, multiplied by the load factor of the new UE ÁL: ÁI ÁL % dI total d , ÁI % dI total d ÁL , ÁI % P N ð1 À Þ 2 ÁL , ÁI % I total 1 À  ÁL , ð9:13Þ The second uplink power increase estimation method is based on the integration method, in which the derivative of interference with respect to the load factor is integrated from the old value of the load factor ( old ¼ ) to the new value of the load factor ( new ¼  þ ÁL)as follows: ÁI ¼ ð I total À old þÁI I total À old dI total , ÁI ¼ ð þÁL  P N ð1 À Þ 2 d , ÁI ¼ P N 1 À  À ÁL À P N 1 À  , ÁI ¼ ÁL 1 À  À ÁL Á P N 1 À  , ÁI ¼ I total 1 À  À ÁL ÁL ð9:14Þ In Equations (9.13) and (9.14) the load factor of the new UE ÁL is the estimated load factor of the new connection and can be obtained as Á L ¼ 1 1 þ W  Á E b =N 0 Á R ð9:15Þ where W is the chip rate, R is the bit rate of the new UE, E b =N 0 is the assumed E b =N 0 of the new connection and  is the assumed voice activity of the new connection. The downlink admission cont rol strategy is the same as in the uplink, i.e. the UE is admitted if the new total downlink transmission power does not exceed the predefined target value: P total À old þ ÁP total > P threshold ð9:16Þ 266 WCDMA for UMTS The threshold value P threshold is set by radio network planning. Notice that ÁP total both includes the power of the new UE requesting capacity and the additional power rise of the existing UEs in the system due to the additional interference contributed by the new UE. The load increase ÁP total in the downlink can be estimated based on a priori knowledge of the required E b =N 0 , the requested bit rat e, and the pilot report from the UE. The pilot report implicitly provides information on the path loss towards the new UE as well as the interference level experienced by the UE. 9.5.3 Throughput-Based Admission Control Strategy In the throughput-based admission control strategy, the new requesting UE is admitted into the radio access network if  UL þ ÁL UL À threshold ð9:17Þ and the same in downlink:  DL þ ÁL < DL À threshold ð9:18Þ where  UL and  DL are the uplink and downlink load factors before the admittance of the new connection and are estimated as shown in Section 9.4. The load factor of the new UE ÁL is calculated as in Equation (9.15). Finally, we need to note that different admission control strategies can be used in the uplink and in the downlink. 9.6 Load Control (Congestion Control) One important task of the RRM functionality is to ensure that the system is not overloaded and remains stable. If the system is properly planned, and the admission control and packets scheduler work sufficiently well, overload situations should be exceptional. If overload is encountered, however, the load control functionality returns the system quickly and controllably back to the targeted load, which is defined by the radio network planning. The possible load control actions in order to reduce load are listed below:  Downlink fast load control: Deny downlink power-up commands received from the UE.  Uplink fast load control: Reduce the uplink E b =N 0 target used by the uplink fast power control.  Reduce the throughput of packet data traffic.  Handover to another WCDMA carrier.  Handover to GSM.  Decrease bit rates of real time UEs, e.g. AMR speech codec.  Drop low priority calls in a controlled fashion. The first two in this list are fast actions that are carried out within a Node B. These actions can take place within one time slot, i.e. with 1.5 kHz frequency, and provide fast Radio Resource Management 267 prioritisation of the different services. The instantaneous frame error rate of the non-delay- sensitive connections can be allowed to increase in order to maintain the quality of those services that cannot tolerate retransmission. These actions only cause increased delay of packet data services while the quality of the conversational services, such as speech and video telephony, is maintained. The other load control actions are typically slower. Packet traffic is reduced by the packet scheduler: see Chapter 10. Inter-frequency and inter-system handovers can also be used as load balancing and load control algorithms and they were described in this chapter. One example of a real time connection whose bit rate can be decreased by the radio access network is Adaptive Multirate (AMR) speech codec: for further information see Section 2.2. References [1] Sipila ¨ , K., Laiho-Steffens, J., Wacker, A. and Ja ¨ sberg, M., ‘Modelling the Impact of the Fast Power Control on the WCDMA Uplink’, Proceedings of VTC’99 Spring, Houston, TX, 16–19 May 1999, pp. 1266–1270. [2] 3GPP TS 25.101 UE Radio Transmission and Reception (FDD). [3] Salonaho, O. and Laakso, J., ‘Flexible Power Allocation for Physical Control Channel in Wideband CDMA’, Proceedings of VTC’99 Spring, Houston, TX, 16–19 May 1999, pp. 1455–1458. [4] Sampath, A., Kumar, P. and Holtzman, J., ‘On Setting Reverse Link Target SIR in a CDMA System’, Proceedings of VTC’97, Arizona, 4–7 May 1997, Vol. 2, pp. 929 –933. [5] 3GPP, Technical Specification Group RAN, Working Group 2 (WG2), ‘Radio Resource Manage- ment Strategies’, TR 25.922. [6] 3GPP TS 25.133 ‘Requirements for Support of Radio Resource Management (FDD)’. [7] Hiltunen, K., Binucci, N. and Bergstro ¨ m, J., ’Comparison Between the Periodic and Event- Triggered Intra-Frequency Handover Measurement Reporting in WCDMA’, Proceedings of IEEE WCNC 2000, Chicago, 23–28 September 2000. [8] Shapira, J. and Padovani, R., ‘Spatial Topology and Dynamics in CDMA Cellular Radio’, Proceedings of 42nd IEEE VTS Conference, Denver, CO, May 1992, pp. 213–216. [9] Shapira, J., ‘Microcell Engineering in CDMA Cellular Networks’, IEEE Transactions on Vehicular Technology, Vol. 43, No. 4, November 1994, pp. 817–825. [10] Dahlman, E., Knutsson, J., Ovesjo ¨ , F., Persson, M. and Roobol, C., ‘WCDMA – The Radio Interface for Future Mobile Multimedia Communications’, IEEE Transactions on Vehicular Technology, Vol. 47, No. 4, November 1998, pp. 1105–1118. [11] Huang, C. and Yates, R., ‘Call Admission in Power Controlled CDMA Systems’, Proceedings of VTC’96, Atlanta, GA, May 1996, pp. 1665–1669. [12] Knutsson, J., Butovitsch, T., Persson, M. and Yates, R., ‘Evaluation of Admission Control Algorithms for CDMA System in a Manhattan Environment’, Proceedings of 2nd CDMA International Conference, CIC ’97, Seoul, South Korea, October 1997, pp. 414–418. [13] Knutsson, J., Butovitsch, P., Persson, M. and Yates, R., ‘Downlink Admission Control Strategies for CDMA Systems in a Manhattan Environment’, Proceedings of VTC’98, Ottawa, Canada, May 1998, pp. 1453–1457. [14] Liu, Z. and Zarki, M. ‘SIR Based Call Admission Control for DS-CDMA Cellular System,’ IEEE Journal on Selected Areas in Communications, Vol. 12, 1994, pp. 638–644. [15] Holma, H. and Laakso, J., ‘Uplink Admission Control and Soft Capacity with MUD in CDMA’, Proceedings of VTC’99 Fall, Amsterdam, Netherlands, 19–22 September 1999, pp. 431–435. 268 WCDMA for UMTS [...]... Pinging www.nokia.com [1 47. 243.3 .73 ] with 32 bytes of data: Reply from 1 47. 243.3 .73 : bytes = 32 time = 301ms TTL = 242 Reply from 1 47. 243.3 .73 : bytes= 32 time = 290ms TTL = 242 Reply from 1 47. 243.3 .73 : bytes = 32 time = 300ms TTL = 242 Reply from 1 47. 243.3 .73 : bytes = 32 time = 241ms TTL = 242 Reply from 1 47. 243.3 .73 : bytes = 32 time = 290ms TTL = 242 Reply from 1 47. 243.3 .73 : bytes = 32 time = 281ms... = 242 Reply from 1 47. 243.3 .73 : bytes = 32 time = 260ms TTL = 242 Reply from 1 47. 243.3 .73 : bytes = 32 time = 281ms TTL = 242 Reply from 1 47. 243.3 .73 : bytes = 32 time = 300ms TTL = 242 Reply from 1 47. 243.3 .73 : bytes = 32 time = 311ms TTL = 242 Reply from 1 47. 243.3 .73 : bytes = 32 time = 290ms TTL = 242 Reply from 1 47. 243.3 .73 : bytes = 32 time = 300ms TTL = 242 Reply from 1 47. 243.3 .73 : bytes = 32 time... protocol stack for web browsing application is shown in Figure 10.3 It can be seen that the medium access control (MAC), the radio link control (RLC) and the packet data convergence protocol (PDCP) layer are terminated in the RNC, while the Internet WCDMA for UMTS, third edition Edited by Harri Holma and Antti Toskala # 2004 John Wiley & Sons, Ltd ISBN: 0- 470 - 870 96-6 WCDMA for UMTS 270 Example service... bytes of data: Reply from 1 47. 243.3 .73 : bytes = 128 time = 1121ms TTL = 242 Reply from 1 47. 243.3 .73 : bytes = 128 time = 221ms TTL = 242 Reply from 1 47. 243.3 .73 : bytes = 128 time = 190ms TTL = 242 DCH allocation 900 ms (=1121 − 221 ms) Reply from 1 47. 243.3 .73 : bytes = 128 time = 200ms TTL = 242 Reply from 1 47. 243.3 .73 : bytes = 128 time = 190ms TTL = 242 Reply from 1 47. 243.3 .73 : bytes = 128 time = 200ms... because WCDMA for UMTS 292 8.0 Effective Eb/N0 [dB] 7. 5 7. 0 6.5 6.0 5.5 5.0 0% 10% 20% 30% 40% 50% BLER Figure 10.30 BLER vs effective Eb =N0 (lowest number ¼ highest capacity) those resources must be reserved for a longer time for the retransmissions The BLER level of 1–10 % is generally assumed for packet data in this book 10.5.2 System Level Performance In this section, the system level performance... TTI Figure 10.10 WCDMA round trip time for small packets Pinging 213.161.41. 37 with 32 bytes of data: Reply from 213.161.41. 37: bytes = 32 time = 201ms TTL = 255 Reply from 213.161.41. 37: bytes = 32 time = 160ms TTL = 255 Reply from 213.161.41. 37: bytes = 32 time = 191ms TTL = 255 Reply from 213.161.41. 37: bytes = 32 time = 190ms TTL = 255 Reply from 213.161.41. 37: bytes = 32 time = 171 ms TTL = 255 Reply... 213.161.41. 37: bytes = 32 time = 180ms TTL = 255 Reply from 213.161.41. 37: bytes = 32 time = 321ms TTL = 255 RLC retransmission Reply from 213.161.41. 37: bytes = 32 time = 170 ms TTL = 255 Reply from 213.161.41. 37: bytes = 32 time = 160ms TTL = 255 Reply from 213.161.41. 37: bytes = 32 time = 180ms TTL = 255 Reply from 213.161.41. 37: bytes = 32 time = 170 ms TTL = 255 Reply from 213.161.41. 37: bytes =... Third generation networks improve the packet data performance compared to second generation systems and bring it closer to fixed line performance Some enhancements have been considered to optimise TCP, as described in [2] Some of those solutions may be beneficial for improving the third generation packet data performance as well These potential TCP optimisation solutions can be classified into WCDMA for. .. such as 32 kbps, for all users and then allocate the remaining capacity for the high priority users to increase their bit rate Packets arriving with different QoS requirements Scarce radio resources shared according to QoS Figure 10.26 QoS priorities are targeted for efficient radio utilisation QoS parameters can also be used to differentiate the services for performance monitoring If, for example, MMS,... except for when a block error occurs in the air interface and an RLC retransmission takes place, in which case the round trip time is increased to 320 ms Since WCDMA round trip time is similar to the fixed Internet round trip time that can be seen for inter-continental connections or for dial-up connections, we may expect that those applications that are designed for fixed Internet Packet Scheduling 277 WCDMA . seen for inter-continental connections or for dial-up connections, we may expect that those applications that are designed for fixed Internet Figure 10.9. Round trip time definition 276 WCDMA for UMTS typically. terminated in the RNC, while the Internet WCDMA for UMTS, third edition. Edited by Harri Holma and Antti Toskala # 2004 John Wiley & Sons, Ltd ISBN: 0- 470 - 870 96-6 Protocol (IP), TCP and hypertext. DATA- K packet Figure 10 .7. Redundancy elimination 274 WCDMA for UMTS  Split TCP. Splitting the TCP connections in to two legs (one leg for the wireless domain and one leg for the fixed Internet) allows a more