Optimizing LTE Network Performance with Tower Mounted Amplifiers

WHITE PApER Optimizing LTE Network Performance with Tower Mounted Amplifiers 1

Table of Contents 1. Overview... 3 2. Background... 5 3. enodeb Receiver Performance... 5 4. Cell Site Performance... 8 5. Increasing Network Coverage and Capacity... 9 6. Key TMA Performance Parameters... 10 7. Effect of TMA Installation on Site Performance... 13 8. Summary... 14 Authored by Mike Eddy Vice President, Cell Site Optimization Westell Technologies 2

OVERVIEW Demand for wireless capacity continues to grow, driven by increasing wireless data usage. In 2012 global mobile data traffic grew 70%, with video traffic increasing to more than 50% of the total traffic. These trends are expected to continue for the next several years (see figure 1). Figure 1. Wireless data forecast, 2013-2017 With this dramatic increase in wireless traffic, dominated by wireless data, there is an ever increasing demand for more network capacity. However, wireless data is much less profitable than traditional voice, so network infrastructure and transport technology must be used as efficiently as possible. Figure 2. Change in dynamics for wireless revenue generation and traffic, as the wireless user transitions to smartphones and uses wireless devices for internet browsing and video streaming 3

LTE (Long Term Evolution) is the wireless technology being deployed globally, led by deployments in the US. LTE was developed based upon a priority for wireless data transport, efficient spectrum utilization and a system architecture designed from the beginning to be low cost and flat, so as to minimize nodes and connections. Gateway GPRS Support Node System Architecture Evolution Gateway Serving GPRS Support Node Mobility Management Entity Control Plane User Plane Radio Network Controller Node B enode B Release 6 HSPA Release 8 LTE Figure 3. HSPA (Release 6) and LTE (Release 8) Network architecture Consequently the enodeb takes more responsibility for the radio control functions, resulting in less delay in the decision process. Figure 4. Overall E-UTRAN architecture 4

The US is leading the way in deployment of LTE and after the initial deployment, optimization of coverage and capacity will follow. In addition, because of the favorable economics, users will be encouraged to migrate from 3G/CDMA as quickly as possible. BACKGROUND For any wireless system, the two critical components are the cell site, comprising enodeb and antennas, and the mobile devices (UE, User Equipment). The cell site is at a fixed location, while the mobile can vary location. Propagation losses will be similar for both transmit and receive. The difference between the power emitted from the mobile compared to the enodeb dominates in determining the difference in coverage limits of the wireless connection. Receive/Uplink Transmit/Downlink Figure 5. Transmit and Receive of RF signals to and from a mobile device. The mobile phone or laptop is powered by battery, so the maximum transmit power is low (200mW), while the maximum power from the enodeb can be as high as 60W. Therefore the weak link is usually the link from the mobile to the cell site (Receive or Uplink). enodeb RECEIVER PERFORMANCE Key performance metrics for any receiver are, Receiver Sensitivity and Dynamic Range. Receiver Sensitivity Receiver Sensitivity (the ability to detect a signal above the noise floor at a certain Bit Error Rate (BER)) is a function of three parameters. 1. Ambient/Thermal Noise Power Ambient or Thermal Noise Power is a measure of the background/random noise generated by all matter. The Noise Power (N 0 ) is : N o = ktb (1) 5

where k is Boltzman s Constant, 1.38 * 10-23 Joules/Kelvin T is the Absolute Temperature, K B is the Bandwidth, Hz In the case of LTE, each resource block bandwidth is 180kHz. So, the background noise power is -121dBm. 2. Signal to Noise Ratio Signal to Noise Ratio (SNR) is a measure of the strength that a received signal must be above the noise floor in order to maintain a certain Bit Error Rate (BER), or in the case of data, Block Error Rate (BLER). Adaptive Modulation matches SNR with modulation and coding scheme (MCS) (and hence data rate). For example, for a particular LTE MCS 3.5dB SNR is required. Thus a received signal must be 3.5dB above the background noise in order to use this MCS. 3. Noise Figure Noise Figure, NF, is a measure of the noise added by components within the receiver chain. Typically the NF of a receiver is 4dB, but is very dependent on the design of the receiver. Thus the noise floor is the ambient noise + noise figure of the receiver (N receiver ), or -121dBm + 4dBm = -117dBm. The receiver sensitivity is the sum of the ambient noise, noise figure of enodeb and other components in the receiver chain, such as feeder cables, jumper cables, lightning protection, duplexer/diplexers (typically 3dB but frequency dependent) and the required SNR. Sensitivity = N o + N receiver + SNR (2) Sensitivity = -121 +4 (from enodeb NF) +3 (from other Rx components NF) +3.5 (from SNR) = -110.5dBm -110.5-114 Receiver Sensitivity (dbm) -121 Receive Bandwidth Frequency Figure 6. enodeb sensitivity for the example described above. If the mobile is at the cell edge, it may have insufficient power to maintain the required 3.5dB SNR for this datarate. 6

Dynamic Range A receiver s dynamic range is the ability of a receiver to handle a range of signal strengths from the weakest to the strongest. The low end of the dynamic range is set by the receiver noise floor. The high end of the dynamic range is set by the ability of the receiver to handle high power signals. Output In Compression Linear Region Below Detection/Noise Floor Input Figure 7. Depiction of Dynamic range. Dynamic range is the linear portion of input to output signal strength that can be detected. enodeb receiver design balances the trade-off between noise figure (sensitivity) and distortion(dynamic range) in the receiver chain by optimizing both the size and location of the gain elements in the receiver. The need to amplify the signal as early as possible is shown in the cascaded noise factor equation (equation 3). F Total = F 1 F2 1 F3 1 Fn 1 + + +... + G G G G G... G 1 1 2 1 2 n 1 (3) Where G is the gain of each element, and F is the noise factor of an element in the receive chain and is related to noise figure, N, by : N = 10 log 10 (F) (4) Noise Factor is linear, while Noise Figure has dimensions of db. From (3), the more gain at the front of the receiver chain the better the noise factor, and hence noise figure. However, additional gain at the front of the receiver will also increase the size of signals passing through the components of the receiver chain, which, due to their non-linear characteristics will generate intermodulation noise, and potentially saturate elements in the receiver, reducing the dynamic range. Thus the best noise figure receiver design requires maximum gain at the front end of the receiver, but with two critical provisions: 1) Out-of-band signals that produce IMD noise must be eliminated by filtering, and 2) in a high signal environment the additional gain should not saturate any component in the receiver, in order to take full advantage of the benefit. Thus, a high gain, low noise figure, high rejection front-end with provision to vary the gain. 7

Note that the magnitude of the improvement in receiver performance by adding a high gain element at the front end of a receiver is not simply the gain of the added element (since noise is also amplified), but is related to the gain and noise figure by equation 3. CELL SITE PERFORMANCE For any wireless technology that can allocate all of the mobile or enodeb power to an individual user, in a rural environment (low density of users), the receive path will limit the coverage area. For example, consider a 20W (43dBm) power amplifier emitting from enodeb, and the maximum power possible from a mobile is 0.2W (23dBm). Then the link imbalance is 14dB in favor of the transmit path. Transmit Path Receive Path Transmit 43 dbm 23 dbm power Diversity 0 3 db Receiver (7) db (4) db noise figure Imbalance 14 db Table 1: Path analysis for limits assuming all transmit power is allocated to an individual user (low density of users). Fundamentally, high power amplifiers can be accommodated at a cell site, whereas the size and power constraints of a mobile unit such as a handset, limit the power possible. As the user density increases (urban environments) then the enodeb transmit power is divided amongst the users, whereas the maximum power from the mobile remains the same, at 23dBm. Transmit Path Receive Path Transmit 32 dbm 23 dbm power/user Diversity 0 3 db Receiver (7) db (4) db noise figure Imbalance 3 db Table 2: Path analysis for limits assuming all transmit power is allocated to an individual user (high density of users). Noise and Signal-to-Noise Before we spend more time on characterizing TMA performance metrics, it is worthwhile providing more detail on the potential noise sources that affect signalto-noise-ratio. As described earlier, the noise in SNR includes both background noise and noise added by components in the receiver chain. It also includes noise added from interfering signals (hence sometimes called Signal to Interference and Noise Ratio, SINR). So, the total noise is: or, N Total = N o + N receiver + N interference (5) N Total = F Total N o (6) The noise generated from interference can come from co-channel interference interference from signals using the same frequency, say from a neighboring sector or from intermodulation distortion (IMD), from interfering signals that are outside the receive band, but generate IMD noise in-band due to the non-linearity of components in the receiver. TMAs will protect against potential out-of-band interferors, while also reducing the receiver-added noise. Signal Interference Receiver Background Signal Interference Receiver Background Figure 8. Schematic showing reduced receiver noise on noise floor. In this case the interference noise remains unchanged. However, for LTE, in high user environments, co-channel interference, from other mobiles using the same frequency in a different sector, can dominate. The magnitude of the co-channel interference depends upon the cell site details, user distribution as well as the network parameters, in particular, the mobile power control parameters. 8

Although the imbalance is reduced as the number of users increases, the receive path remains the limiting link. In addition, as the power in the power amplifiers in the enodeb are increased, this imbalance will be further exacerbated. Thus, in order to improve cell site performance the sensitivity of the receive path must be increased. So how can we improve the receive performance (receiver sensitivity)? INCREASING NETWORK COVERAGE AND CAPACITY There are a limited number of techniques to enhance receive path performance. Each method improves SNR, either by increasing the signal strength or decreasing the noise. A fundamental assumption is that the spectrum available is limited, as more spectrum enables more capacity. So, possible means to improve receive path performance are: 1. Increase the number of receive antennas. Diversity. Most base stations use two receive antennas for each sector. Two antennas are used to mitigate fading, which arises from interference nulls, when multipath signals reflected from surfaces such as buildings or roads reduce the signal strength at the antenna (due to destructive interference), as well as adding signals from the two receive paths. This basic concept can be extended, adding antennas, and ultimately beam forming to focus energy where the mobile traffic is located. This gain will improve the receiver sensitivity and is the foundation of spatially resolving users and smart antennas. However, it is costly, requiring antennas, cables and receivers. Also, it is complex in integration and execution, particularly for fast moving mobiles. 2. Increase power at the remote terminal. This can be achieved at: a. Mobile A higher power handset or mobile device, will improve the reception of the signal at the base station. A call is dropped when the maximum power from the handset is reached. Consequently, with more available power the handset can move further from the tower before reaching the maximum threshold. However, the maximum power in a mobile device will always be constrained by the battery power available and emissions from the power amplifier causing interference. Early mobile devices were capable of 1W of power, but a typical handset available today will only emit ~200mW. Efficient use of the available power is also important. Hence control messages are sent to the mobile device to control the power arriving at the cell site, so as to minimize interference. This basic concept of increasing power at a remote terminal can be applied as a two-step process, or booster implementation, where a remote unit boosts the mobile signal prior to re-transmission to the base station. This forms the basis of repeater and remote RF network architectures. b. Repeater/remote RF unit Rather than produce more power from the terminal, a more feasible approach is to locate a remote booster close to traffic concentrations. This remote unit has the capability of receiving the signal from the terminals and then boosting the signal back to the base station. The remote unit can now have a much larger power amplifier for re-transmission to the base station/host. This network architecture has attractions, but critical to overall performance is the transport mechanism back to the BTS and the impact on the RF front-end. Typically, significant gain in the remote is required to overcome the losses associated with the transport and minimizing the additional noise generated at the BTS by the remote. This extra gain will require improved filter performance in order to minimize IMD generation. 9

3. Add amplification of the signal prior to the enodeb. This is the foundation of tower mounted amplifiers. From (1), by adding low noise gain on the Rx path, prior to losses from cables and the enodeb the noise figure, and hence receiver sensitivity can be dramatically improved. In order to take full advantage of the potential benefits, the TMA must be designed with several key parameters in mind. KEY TMA PERFORMANCE PARAMETERS Tower mounted amplifiers consist of an amplifier protected by filters so that the signals that are amplified are kept below a threshold so as not to produce any unwanted IMD noise, or that the amplifier is driven into compression. The configuration for these amplifier systems depends upon the configuration of the cabling and antennas at the tower. However, most are dual-duplexed (figure 5), because the Tx and Rx signals are on the same coax cable. Pre LNA r ecei ve filter LNA with opti onal bypass Post LNA receive filter From Antenna Central c ontrol : monitoring, al arm and communicati ons To BTS Low loss transmit filter Lightning pr otecti on Figure 9: Simplified TMA schematic for dual-duplexed configuration. So within the TMA, the Tx and Rx signals are separated, the Rx signal amplified and then Tx and Rx recombined. The important parameters for a TMA are: 1. TMA Gain The total gain added to the receive path from a TMA should have the lowest noise figure possible. The total gain added from the TMA is a trade-off between reduced receive path noise figure and reduced dynamic range. Note that there is a diminishing return to the improved noise figure of the receive path as more gain is added by the TMA (figure 10). Most of the benefit (>90%) can be realized with 12-13dB additional gain. 10

7 6 5 Rx NF Improvement 4 3 2 1 0-1 0 5 10 15 20 TMA Gain (db) Figure 10: Receiver noise figure improvement as a function of TMA gain. Consequently, by deploying a TMA with 13dB gain on a 150ft tower, almost 6dB of improved receiver sensitivity can be realized. 2. TMA Noise Figure A TMA is deployed close to the antenna, therefore the most significant contribution to the overall receive sensitivity is the noise figure of the TMA (see (3)). Improvement in TMA noise figure is directly related to improved coverage of a cell-site. Baseline - No TMA db Linear Jumper Cable NF F1 0.20 1.0 Jumper Cable Gain G1-0.20 1.0 Feeder Cable Loss F2 3.00 2.0 Feeder Cable Gain G2-3.00 0.5 Jumper Cable NF F3 0.50 1.1 Jumper Cable Gain G3-0.50 0.9 enodeb NF F4 4.00 2.5 SYSTEM NF 7.70 5.9 With Westell TMA db Linear Jumper Cable NF F1 0.10 1.0 Jumper Cable Gain G1-0.10 1.0 Westell TMA NF F2 1.10 1.3 Westell TMA Gain G2 13.00 20.0 Jumper Cable NF F3 0.10 1.0 Jumper Cable Gain G3-0.10 1.0 Feeder Cable Loss F4 3.00 2.0 Feeder Cable Gain G4-3.00 0.5 Jumper Cable NF F5 0.50 1.1 Jumper Cable Gain G5-0.50 0.9 enodeb NF F6 4.00 2.5 SYSTEM NF 1.94 1.6 Thus, for a site with 3dB main feeder cable loss, installing a TMA with 13dB gain and 1.1dB noise figure will improve the Rx system noise figure by 5.76dB (see tables). This corresponds to an increase in cell site radius of 30% or 100% increase in coverage area. 11

With TMA No TMA Figure 11: Expected increase in cell-edge coverage when deploying a TMA on a typical cell site tower. For LTE, because each MCS is determined based upon SNR, there is, effectively, an improved coverage area, similar to the cell-edge coverage, for each MCS (data throughput). Figure 12: Coverage area for a particular datarate, based upon MCS selection. Each coverage area will increase with the addition of a TMA at the cell site. Depending upon the SNR at the enodeb, the benefit of the improved receiver sensitivity can also be observed in lower mobile transmit power. This depends upon the SNR and the mobile power control parameter settings. LTE has both open and closed loop power control defined in general power control equation (7). P = min {P max, P 0 + 10log M + α.pl + δ MCS + ƒ ( i ) } (7) where P max is the maximum UE power P 0 is the power in a single resource block, broadcast to the UE and used for intial power setting M is the number of resource blocks α is a cell specific parameter that enables use of fractional power control, broadcast to the UE for initial power setting PL is the estimated downlink pathloss calculated at the UE, used for initial power setting δ MCS is an MCS dependent offset that is UE specific ƒ ( i ) is a function that can be relative, cumulative or absolute corrections to the UE power 12

The initial open loop path loss, PL, is estimated using the transmit reference symbols, measured at the UE. The closed loop component is contained in i. i = (SINR Target - SINR Est. ) (8) At the enodeb, SINR is measured and the UE transmit power adjusted to meet the Target SINR. Adjustments are sent to the UE via the Transmit Power Control (TPC) signal. Optimization of the performance improvements due to the TMA is dependent on how the closed loop power control is implemented. 3. Rejection of interfering signals and linearity of the low noise amplifier Large out-of-band signals will produce intermodulation distortion, or even blocking, in the LNA if the signals are large. This is particularly the case as more networks are deployed and traffic increases, forcing high outof-band signals adjacent to receiver frequencies. However, antenna isolation (more than 30dB) typically reduces the signals from neighboring transmit antennas so that the dominant issue is isolation of transmit and receive within the dual duplex configuration of the TMA. For two transmit tones, each with power of 40W, the rejection required is more that 85dB. This accounts also for the higher power peaks associated with the high peak-to-average ratio of OFDMA signals. Other Considerations Beyond the RF parameters, TMAs are an investment in the network, and therefore, must: Be easily installed Be highly reliable (more than 1million hours MTBF, less than 0.5% annual failure rate) Include lightning protection Allow for communication via coax, preferably using AISG protocol Be temperature stable No TMA With TMA Receive Path Noise Figure 7.70dB 1.94dB Improvement with TMA 5.7dB EFFECT OF TMA INSTALLATION ON SITE PERFORMANCE Improved receiver sensitivity increases cell site coverage. This will be particularly important as LTE wireless users become more mobile, for example, as Voice over LTE (VoLTE) is implemented. Users that were beyond the coverage of a cell site can now be covered, increasing traffic for the site. Key performance metrics, such as hand-overs, dropped calls and blocked calls will improve. The addition of 13dB gain from a TMA will increase the noise floor, as measured at the enodeb, by the gain of the TMA. So for a noise floor of -117dBm prior to install of a TMA the noise floor will increase to ~-104dBm. Since the high end of the dynamic range does not change, the dynamic range of the receiver will be reduced. 13

This is not an issue for receivers with high dynamic range (55dB and higher), or if the only signals in-band are tightly power controlled. Where this is an issue is in the case of a low dynamic range receiver and high in-band signals (from high traffic or PIM). The best solution, if receiver components are being saturated from high in-band signals once a TMA is installed, is to reduce the total receiver gain within the enodeb. This is a better solution than reducing the TMA gain, because the high front-end, low noise figure gain is maintained (see comparison below) With 8dB Gain removed from TMA db Linear Jumper Cable NF F1 0.10 1.0 Jumper Cable Gain G1-0.10 1.0 Westell TMA NF F2 1.92 1.6 Westell TMA Gain G2 5.00 3.2 Jumper Cable NF F3 0.10 1.0 Jumper Cable Gain G3-0.10 1.0 Feeder Cable Loss F4 3.00 2.0 Feeder Cable Gain G4-3.00 0.5 Jumper Cable NF F5 0.50 1.1 Jumper Cable Gain G5-0.50 0.9 enodeb NF F6 4.00 2.5 SYSTEM NF 4.96 3.1 With 8dB Gain removed from enodeb db Linear Jumper Cable NF F1 0.10 1.0 Jumper Cable Gain G1-0.10 1.0 Westell TMA NF F2 1.10 1.3 Westell TMA Gain G2 13.00 20.0 Jumper Cable NF F3 0.10 1.0 Jumper Cable Gain G3-0.10 1.0 Feeder Cable Loss F4 3.00 2.0 Feeder Cable Gain G4-3.00 0.5 Jumper Cable NF F5 0.50 1.1 Jumper Cable Gain G5-0.50 0.9 enodeb NF F6 9.34 8.6 SYSTEM NF 3.57 2.3 Improved Rx Noise Figure 1.38 However, if the enodeb gain cannot be reduced then varying the gain of the TMA is necessary to optimize the trade-off between performance benefit and reduction in dynamic range. Note that in either case adding the TMA improves the site receiver sensitivity. SUMMARY LTE performance is limited by the link from the UE to the cell site. There are only a limited number of ways to improve this link, and therefore, improve the overall site performance. Tower mounted amplifiers represent a proven and cost-effective solution to improve this link. However, because of the susceptibility of LTE to any additional noise, particular attention must be paid to the performance characteristics of the TMA. Westell TMAs offer state-of-the-art RF performance, amplification of customer signals with minimal additional noise, and rejection of any potentially interfering signals. www.westell.com Westell, Inc. 750 N. Commons Drive Aurora, IL 60504 Westell is a registered trademark of Westell, Inc. Availability of features and specifications subject to change without notice. This document is the property of Westell, Inc. and its contents are proprietary and may neither be copied, reproduced nor its contents disclosed to others without prior written agreement from Westell. Copyright 2013 Westell, Inc. All rights reserved. 1307iArA 14