Throughput Performance Insights of LTE Release 8 Malaysia's Perspective

of 6
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
Related Documents
Throughput performance insights of LTE Release 8 Malaysia's perspective.
  Throughput Performance Insights of LTE Release 8: Malaysia's Perspective Abdulaziz M. Ghaleb, David Chieng, Alvin Ting, Ayad Abdulkafi, Wireless Communication Cluster, MIMOS Berhad, Malaysia. {abdulaziz.saleh, ht.chieng, kee.ting, ayadatiyah.abdulkafi} Kim-Chuan Lim, FKeKK, University Teknikal Malaysia Melaka, Heng-Siong Lim, FET, Multimedia University, Malaysia,  Abstract  — LTE wireless mobile broadband networks, in particularly those based on 3GPP Release 8 (Rel. 8) specification, have already made strong inroads into the commercial arena worldwide. In Malaysia, 8 companies have been allocated spectrum in 2.6GHz Band (LTE band class 7). This paper aims to provide some high level insights on the throughput performance of these spectrums. Although a lot of studies have been undertaken with regards to LTE network performances, various degrees of discrepancy still exist in particularly concerning network layer (IP) throughput. A wide array of factors may contribute to these differences, which include differences in methodology adopted, levels of abstraction (or details) used in the simulation model, environment and/or usage scenarios and so on. Using OPNET’s latest LTE library, we study the effects of duplexing scheme (FDD vs. TDD), MCS, channel bandwidth, bearer's type (GBR or non-GBR) for the allocated spectrums. The impact of multiple users’ access on the throughput performance is also analyzed. This work enables us to compare and contrast our findings with the existing studies while providing more accurate views on how the main system level configurations may impact the network layer throughput performance of the emerging LTE networks in Malaysia. Keywords  –   LTE, OPNET, Throughput, Performance. I.   I NTRODUCTION LTE or Long Term Evolution UMTS networks, in particularly those based on 3GPP Rel. 8 specifications, have already made strong inroads into the commercial arena worldwide. According to [1] global LTE subscribership is expected to skyrocket from 9 million in 2011 to more than 560 million in 2016. Global mobile Suppliers Association (GSA) recently announced that 145 LTE commercial networks had been launched in 66 countries (as of January 2013) and that the number is expected to increase to 234 in 83 countries by the end the year. In Malaysia 8 companies have been awarded LTE spectrum at 2.6GHz band. Puncak Semangat received a lion share of 2x20MHz FDD, five companies received 2x10MHz FDD and the two WiMAX mobile broadband providers received 20MHz TDD [2] Table I: LTE Spectrum Recipients in Malaysia[2] Licensee Allocation Type Puncak Semangat 2x20MHz FDD Maxis 2x10MHz FDD Celcom 2x10MHz FDD Digi 2x10MHz FDD U Mobile 2x10MHz FDD Redtone 2x10MHz FDD P1 20MHz TDD YTL 20MHz TDD LTE Rel. 8, though marketed as “4G”, is in fact only “3.9G”. LTE Release 10 (Rel. 10) is the official 4G technology recognized by ITU/IMT-Advanced [3]. LTE and LTE-Advanced are primarily the same technology family, with the “Advanced” label being added to draw attention to the relation between LTE Rel. 10 and ITU/IMT-Advanced. According to Rel. 8 specification, a peak downlink rate of 299.6 Mbps for 4x4 antennas, and 150.8Mbps for 2x2 antennas using 20MHz channel can be provided. As for the uplink, the peak rate of about 75.4Mbps is expected using the same spectrum size. LTE supports scalable carrier bandwidths ranging from 1.4MHz to 20MHz and supports both frequency division (FDD) duplexing and time-division duplexing (TDD) schemes. In terms of latency, LTE promises low latency in data transfer (<5ms in optimal condition), handover as well as connection setup time. Due to that LTE has the ability to manage fast-moving mobiles (between 350 km/h or 500 km/h). Application wise, LTE generally supports all conventional multimedia services with better QoS guarantee including multi-cast and broadcast streaming services. A lot of studies have been undertaken in order to evaluate LTE throughput performance. I. Vukovicv on [4] investigated the performance of TDD LTE RACH (Random Access Channel) and computed the normalized throughput under Poisson load. J. Zhu evaluated the PDSCH link and system level performance and provided a platform to estimate the performance for the different transmission modes [5]. The work done by [6, 7] focused on the downlink control channel design to reduce the control overhead and hence, to improve the throughput performance. To the best of our knowledge, the most comprehensive work with regards to throughput performance, is done by [8]. The authors provided performance analysis of LTE physical layer characteristics by studying the maximum throughput in uplink and downlink for both FDD and TDD using different Modulation and Coding Scheme (MCS) and channel bandwidth sizes. From our investigation, various degrees of discrepancy still exist between the above findings. A wide range of factors may contribute to these differences, such as differences in methodology adopted; levels of abstraction (or level of detail) used in the simulation model, environment and/or usage scenarios and so on. 978-1-4673-2480-9/13/$31.00 ©2013 IEEE 258   This paper aims to provide fresh insights on the network layer (or IP layer) throughput performance of LTE, which represents the PDSCH capacity using a well-known industrial-grade network simulator called OPNET Modeler (ver. 17.5). OPNET Modeler is a powerful simulation tool that offers a comprehensive network simulation platform with many predefined standard node models. Its library offers more than 400 out-of-the-box protocols and vendor device models including IPv6, TCP/UDP, UMTS, WiMAX, LTE and WLAN. Using its latest LTE Rel.8 library, we study the effects of duplexing scheme (FDD vs. TDD), MCS, Channel Bandwidth, Bearer's type (GBR or non-GBR). The impact of multiple users’ access on the throughput performance is also analyzed. The LTE model greatly benefits from the comprehensive higher layer protocols and powerful statistical evaluation tools. The level of details of the LTE model provided by OPNET can be appreciated in the following sections. This work enables us to compare and contrast our findings with the existing studies. Besides, this work provides more accurate views on how the main system level configurations may impact the throughput performance of the emerging LTE networks in Malaysia. The rest of the paper is organized as follows. Section II provides a quick overview on LTE radio access technology. We then describe the simulation configuration and setup in Section III. Different scenarios are studied in Section IV and finally the conclusions are drawn in Section V. II.   B ACKGROUND   O F   L TE   R ADIO   A CCESS The LTE standard was primary published in the first quarter of 2009 as part of the 3GPP Rel. 8 specifications. LTE significantly improves flexibility and overall system performance by utilizing wider spectrum bandwidths wherever and whenever available [9]. Table II summarizes the acronyms used in this paper. Table II: List of Acronyms  DwPTS Downlink Pilot Time Slot  FDD Frequency Division Duplex GBR Guarantee Bit Rate GP Guard Period  HARQ Hybrid Automatic Repeat-Request  ICIC Inter-Cell Interference Coordination  MBMS Multimedia Broadcast Multicast Services  MCS Modulation And Coding Scheme  MIMO Multiple-Input Multiple-Output OFDMA Orthogonal Frequency Division Multiple Access  PBCH Physical Broadcast Channel  PCFICH Physical Control Format Indicator Channel  PDCCH Physical Downlink Control Channel  PDSCH Physical Downlink Shared Channel  PUCCH Physical Uplink Control Channel  PHICH Physical HARQ Indicator Channel  PHY Physical Layer  P-SS Primary Synchronization Signal QCI QoS Class Identifier SC-FDMA Single Carrier Frequency Division Mulitple Access SNR Signal-To-Interference Ratio SRS Sounding Reference Signal S-SS Secondary Synchronization Signal TDD Time Division Duplex UE User Equipment UpPTS Downlink Pilot Time Slot *Due to space limitation, not all the parameters can be listed. LTE uses OFDMA signal for downlink and the SC-FDMA signal for uplink. SC-FDMA is used to enable higher terminal power-amplifier efficiency by providing better peak-to-average power ratio. Spectrum flexibility is a main feature of the LTE radio-access technology. This allows LTE to utilize scalable spectrum (different bandwidth sizes) at difference frequency bands with different characteristics including different duplex schemes such as FDD and TDD. MIMO technique is supported in LTE from its very first release. However, at the point of writing, this feature has yet to be implemented in OPNET Modeler. Channel-Dependent Scheduling and Rate Adaptation are implemented in Rel.8 in order to match to the rapidly varying resource requirements, which are key determinants of the overall system performance. LTE is designed in such a way that each cell can access the entire available spectrum, which subsequently improves the system spectral efficiency. However, one-cell reuse approach causes huge variation in the SNR and thus in the data rates especially at cell edge. Due to that, ICIC is implemented to avoid scheduling transmissions from/to UE at the cell edge at the same time in the adjacent cells [3]. In LTE, fast HARQ with soft combining is used as error control method during retransmission. This technique offers rate adaptation and minimizes the impact of the erroneously received packets [10, 11]. III.   S IMULATION  C ONFIGURATION  A ND  P ARAMETERS   In order to achieve the peak throughout, only one UE is used and only with one type of traffic (default bearer mode) in most scenarios. Only in case study D that the UE is configured with multiple QCIs as well as multiple UEs with multiple QCI to study the effects of QCI type and number of users on the throughput performance.   Table III lists the default parameters unless stated otherwise. 259    Table III: Default parameters. Parameter Settings General Parameters Bearer Type Default Scheduling Mode No Link Adaptation Scheduling request error Disabled Scheduling Info. error Disabled Scheduling Grant error Disabled PDCP Compression Disabled Pathloss Model Free space PHY Frequency band 2.6 GHz Transmission Mode SISO Cyclic Prefix Normal (7 symbols per slot) TDD Channel Index Config 6: UL/DL 5:5 Preamble Format Format 0 ACK-to-NACK error Disabled NACK-to-ACK error Disabled Antenna types Omnidirectional (UE&eNodeB) eNodeB antenna gain 18 dBi eNodeB Max. Tx power 40 Watt eNodeB antenna gain 0 dBi UE Maximum Tx power 200mWatt MCS 1-28 Channel Bandwidth 1.4 to 20MHz Traffic Characteristics Traffic Direction Downlink/Uplink Type of Service Best Effort Protocol UDP IP Packet Size 1500 Bytes Control Signal and Overhead PDCCH 1 OFDM symbol per subframe PCFICH Included in PDCCH PHICH Included in PDCCH P-SS and S-SS 144 REs every 10ms PBCH 288 REs every 10ms Reference signals 8 REs (2 within the PDCCH) PUCCH 2 OFDM symbols/subframe SRS Up to 12 REs Fig. 1: OPNET Simulation Setup Fig. 1 shows the simulation setup in OPNET. The dashed blue line represents the traffic flow direction between the server and the UE. The actual traffic flows through the EPC or the solid lines, which are configured as 1 Gbit/s Ethernet duplex links. Our study focuses only on the radio interface between the eNodeB and the UE. This section represents study cases and discussion of the different scenarios (A to E). Scenarios A to D are based on single UE and single eNodeB, and Scenario E contains multiple users and one eNodeB. FDD transmission mode is used throughout the simulation except for one scenario where FDD and TDD are compared. The results obtained are analyzed, and sub-conclusions are attained from each case study. As mentioned earlier at the point of writing, MIMO configuration is not supported by OPNET. However based on the baseline performance (SISO mode), we can easily estimate the gain that MIMO can offer. IV. S IMULATION S CENARIOS  A.    Effects of Channel Bandwidth Fig. 2 shows the peak uplink and downlink throughput of FDD LTE using one eNodeB and one UE. The results represent the maximum possible throughput attainable by a user (UE). The UE is placed at the distance where the best MCS index = 28 can be supported. Since there is only one base station in this setup, there is no interference experienced between users. The maximum number of RBs, occupied bandwidth and bandwidth utilization for each bandwidth are given in Table IV as follows: Table IV: No. of RBs, Occupied bandwidth and Bandwidth Efficiency andwidth (MHz) 1.4 3 5 10 15 20 o. of RBs 6 15 25 50 75 100 ccupied Bandwidth 1.08 2.7 4.5 9 13.5 18 andwidth Utilization 77% 90% 90% 90% 90% 90% The occupied bandwidth refers to the bandwidth that is actually occupied by the RBs. For the 1.4MHz case, there are 6 RBs with 180KHz each; hence the actual bandwidth is 6x180KHz, which is 1.08MHz. Bandwidth utilization is equal to the ratio between the actual bandwidth and the allocated bandwidth. From Fig. 2, we can observe that the throughput performance is proportional with the channel bandwidth except for the 1.4MHz case. Here, the downlink direction experiences slightly lower throughput than the uplink. This is largely due to the synchronization signals and broadcast signals which are sent at the downlink direction. Hence over the narrow 1.4MHz bandwidth channel, the relative overhead becomes comparatively higher. This is due to the difference in the allocation of control signaling resource. In our simulation PUCCH occupies 28 REs (for all channel bandwidths) while PDCCH is set to one symbol (7REs). This reduces the uplink physical data rate by 10.6% as compared with the downlink. In addition, SRS occupies up to 12 REs while Reference Signal occupies 8 REs, which two of them are included in PDCCH. From Fig. 2, it can be 260   deduced that the 2x10MHz FDD spectrum receipients in Malaysia can only expect up to 43.2Mbps IP layer throughput at the downlink and 37.7Mbps at the uplink using SISO or MISO configuration. The only 2x20MHz license recipient, on the other hand, can enjoy up to 88.5Mbps and 79.3Mbps at the downlink and uplink respectively, which is in fact more than double of the 2x10MHz recipients. 1.4MHz3MHz5MHz10MHz15MHz20MHz102030405060708090100 Bandwidth    L   T   E   T   h  r  o  u  g   h  p  u   t   (   M   b  p  s   )   UplinkDownlink   Fig. 2: Effects of Channel Bandwidth on Throughput (FDD, MCS=28) 20 TDD10 FDD1020304050 Bandwidth (MHz)    L   T   E   T   h  r  o  u  g   h  p  u   t   (   M   b  p  s   )   TDDFDD   Fig. 3: Max Downlink Throughput of 10MHz FDD and 10MHz TDD 5:5 (MCS=28)  B.   TDD versus FDD LTE supports both paired and unpaired spectrum through TDD-LTE. TDD-LTE is different from FDD-LTE in terms of frame structure, scheduling, HARQ and ACK/NACK procedures. 3GPP has specified a special subframe that allows switching between downlink and uplink transmission TDD-LTE. The special subframe contains DwPTS, GP, and UpPTS. GP is required to guarantee that uplink and downlink transmissions do not collide, but larger GP means lesser capacity. For long distance transmission, larger GP is necessary to accommodate larger propagation time. Fig. 3 compares the peak downlink throughput of 10MHz FDD and 20MHz TDD (5:5) when applying the frequency reuse of one (inner cell zone). It can be observed that the downlink throughput 20MHz TDD with 5:5 DL: UL ratio can only reach up to 41Mpbs as opposed to 43.2Mbps in the 10MHz FDD case. The 5.3% difference is largely due to higher overhead incurred in the TDD frame because of the added special subframe. C.    Effects of MCS index LTE eNodeB supports 29 different MCSs with index ranging from 0 to 28. Each MCS is mapped to what is known as transport block size index I TBS ranging from 0 to 26 [12]. I TBS together with the number of RBs determine the transport block size, in bits, that can be transmitted within one TTI. In the downlink direction, MCS with the index 0 - 9 are modulated using QPSK, index 10 - 16 are modulated using 16QAM and the rest are based on 64QAM. The coding rate of each MCS is given in Table V as follows. Table V: MCS indexes and Coding Rates  MCS  Index Coding  Rate  MCS  Index Coding  Rate  MCS  Index Coding  Rate 0 0.16667 10 0.33333 20 0.55556 1 0.2 11 0.35 21 0.6 2 0.23333 12 0.41667 22 0.64444 3 0.26667 13 0.48333 23 0.71111 4 0.33333 14 0.51667 24 0.75556 5 0.4 15 0.58333 25 0.8 6 0.4667 16 0.63333 26 0.84444 7 0.53333 17 0.42222 27 0.88889 8 0.6 18 0.48889 28 1 9 0.66667 19 0.5 In this scenario, we investigate the maximum data rate that each MCS index can support using 20MHz FDD error-free channel. As shown in Fig. 4, the performance is quite linear to the MCS indexes except for MCS 28, which has a relatively higher data rate due to the use of uncoded transmission (coding rate is of 1). MCS 9 and 10, which are modulated QPSK and 16QAM respectively have the same throughput as they are mapped to the same TBS (I TBS =9). The same goes with MCS 16 and 17. Therefore for users subscribing to 2x20MHz FDD network, they will get between 3.22Mbps and 88.5Mbps of IP layer downlink throughput depending on the signal quality they are experiencing as well as how many other subscribers are sharing the downlink channel. Although these results were generated using in SISO mode, the maximum possible throughput will remain the same even we have multiple antennas system at the BS operating in diversity mode. When operating in spatial multiplexing mode e.g. 2x2 MIMO configuration, we can expect around 1.6 or 1.7x gain 261
Related Search
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks