Multi-Cell Interference Coordination in LTE Systems using Beamforming Techniques

Transcription

1 Multi-Cell Interference Coordination in LTE Systems using Beamforming Techniques Sérgio G. Nunes, António Rodrigues Instituto Superior Técnico / Instituto de Telecomunicações Technical University of Lisbon, Lisbon, Portugal Sergio.galveia.nunes@gmail.com, Antonio.rodrigues@lx.it.pt Nuno Cota Instituto Superior de Engenharia de Lisboa Lisbon Polytechnic Institute Lisbon, Portugal ncota@isel.ipl.pt Abstract The main objective of this work was to present the problem of the effect of interference in a LTE (Long-term Evolution) network and propose a solution using Beamforming, mitigating the effects of interference, increasing the quality of network coverage and thus improving the fairness of service to users. Various scenarios are proposed in order to obtain a comparison between a system with and without Beamforming. The results show that in the presence of strong interference, the quality of the coverage of a LTE cell is greatly affected. The use of beamforming allows for an increase of Signal to noise plus interference ratio (SINR) power in cell-edge of around +11 db and this is reflected with an increase of up to +54 Mbps of downlink throughput for the cell-edge user. Spectral efficiency is also greatly improved with results showing a significant increase of up to +3 bits/hz. These results lead to the conclusion that the use of beamforming applied to LTE technology has enormous potential, especially in scenarios where there is a lot of inter-cell interference and therefore these techniques can be considered as a valid alternetive for Inter-cell Interference Coordination (ICIC). Keywords LTE; MIMO; ICIC; Beamforming; QoS I. INTRODUCTION Recently there has been a continuous search for the technological evolution of mobile communications, especially since there are more and more services that demand an increased performance and robustness, even in high mobility environments. To improve the overall capacity of LTE, multiple multipleinput multiple-output (MIMO) techniques have been proposed, allowing users to reach a high spectral efficiency through an intelligent allocation of electromagnetic spectrum. MIMO systems allow for the creation of multiple streams of information that in theory would lead to a multiplication of the capacity of the system using the same number of resources. To increase spectral efficiency even further, the same frequency bands are re-used in multiple locations. This creates the need for flexible spectrum allocation, offering a high scalability factor to the system. The use of Orthogonal Frequency Division Multiplexing (OFDM) and Spatial Division Multiple Access (SDMA) multiplexing techniques are considered as they offer an efficient way of sharing resources, thus increasing spectral efficiency. In order to further improve this system, the introduction of Beamforming techniques is proposed. These techniques allow the creation of very high directive beams in order to substantially improve the coverage of a user in any given time. Beamforming will also allow for a substantial reduction on interference caused to the other users in the cluster, by directing the minimal lobes of radiation to those users. II. THEORETICAL MODELS A. Multiple Access and Multiplexing techniques Reaching the peak data rate of over 100 Mbps is impossible using the current 3rd generation mobile network, which uses Wideband Coded Division Multiple Access (WCDMA). LTE uses OFDMA in the Downlink in order to achieve high peak throughputs with large bandwidths. The scheduling in the frequency domain allows for a mitigation of the interference, increasing spectral efficiency. OFDMA creates several channels with flexible spectral dimensions, varying from 1.4 MHz to 20 MHz of bandwidth. Using a pure OFDMA in the Uplink compromises the efficient use of power in the UE because of the high PAR, drastically reducing the battery life of the equipment. The multiple access technique used is Single-carrier FDMA, which is similar to the OFDMA but presents an advantage of 2 to 6 db in the PAR in relation to pure OFDMA. In a wireless network with high capacity it s crucial to choose the correct modulation. In OFDMA the spectrum is divided in several sub-carriers which are orthogonal, greatly reducing the ISI. Each sub-carrier is modulated independently and can transport a stream of different information to a different user. B. LTE The accelerated growth in the telecommunications led to a change in the way clients access the internet. This in turn is reflected in a decrease of revenue and an exponential increase in traffic.

2 The optimization of packet switch oriented networks, as well as the incessant increase of throughput, mostly because of HD content on the web, led the 3GPP to create the standards to the next generation of the mobile network technology, LTE. LTE presents a transversal architecture which is able to operate simultaneously with the previous generations. The main functionalities of LTE are: Peak Downlink throughput of up to 326 Mbps in a 20 MHz bandwidth system. Peak Uplink throughput of up to 86.4 Mbps in a 20 MHz bandwidth system. Operability in both FDD and TDD modes Scalable bandwidths up to 20 MHz including 1.4, 3, 5, 10, 15 AND 20 MHz Substantial reduction of RTT latency to 10 ms, as well as a reduction to less than 100 ms for activation time. Self-optimization network functionalities, according to the operator s control and preferences, optimizing network planning and reducing implementation costs. The LTE network is a bit different from the previous generation especially because it centralizes all the complex processing to the Base Station interfaces (enbs). The network is also oriented towards packet switching. The radio interface used in LTE divides the radio resources in blocks, called Resource Blocks, which are then distributed between the users according to their coverage quality. C. Link Budget The Link Budget is used in a telecommunications system to calculate all the gains, losses and powers between the transmitter and the receiver, through the propagation path (free space, fiber, cable, etc). This calculation bears in mind the attenuation due to propagation, as well as all the margins needed to estimate the quality of the signal received. 1) Capacity In order to calculate the capacity of an LTE system we must first calculate the number of Resource blocks available for distribution between the users. NRB is the number of RB. Nu is the number of users. NRB/user is the number of RB per user. Then we calculate the throughput offered to each user: (1) Rb/u is the throughput needed for each user (on the physical layer). N symbols/sub-frame is the number of OFDM symbols per sub-frame (depends on the size of the Cyclic Prefix). M is the modulation order (QPSK, 16-QAM, 64- QAM). N streams is the number of streams to consider in MIMO. TTTI is 1 ms in LTE. Peak data rates are then calculated using: (3) N bits is the number of bits used in the modulation. LB is the bandwidth. N subcarrier is the number of sub-carriers. N OFDM symbols per sub-frame is the number of OFDM symbols per sub-frame. Tf is the frame duration, 1ms. 2) Coverage The coverage of a given LTE cell is given by the radius of that cell: (4) P TX-subcarrier[dBm] is the transmited power in each subcarrier. P r-subcarrier[dbm] is the sensitivity needed to reach a certain throughput. D. MIMO In order to surpass the performance obtained with the previous technologies, there have been many proposed solutions using Multiple-Input Multiple-Output (MIMO) techniques which explore the characteristics of multipath in radio transmission. The multipath scenario allows for the establishment of multiple channels of communication between the transmitter and the receiver. The increase in throughput in a MIMO system is proportional to the number of transmit antennas, being limited by the spatial layers created. In a 2x2 MIMO system it is then theoretically possible to reach the double of performance than in a 1x1 system. In LTE there are several transmission modes and the most used are defined in the consequent Table I. (2)

3 TABLE I. TRANSMISSION MODES areas. Each area uses its own frequency and bandwidth and therefore produces no interference to neighbor cells and these frequencies can be re-used if the cells using the same frequencies are far enough from each other. This helps mitigate the effect of interference but has a big impact on spectral efficiency. E. Adaptive Modulation Coding In cellular communications the quality of the received signal depends on the quality of the Serving Cell channel, the level of interference from other cells and the noise level. To optimize the capacity and coverage of LTE, the transmitter must react to changes in the Signal-to-interference-plus-noise ratio (SINR) of the receiver and adapt the distribution of resources according to that SINR. This is done through Adaptive Modulation and Coding (AMC) which allows for the modification of the modulation and codification of a transmission in a quick and efficient way. In an LTE transmission the enb selects a Modulation Coding Scheme (MCS) according to the Channel Quality Indicator (CQI). The CQI is derived from the SINR in the User Equipment (UE) and then sent to the enb via the uplink signaling channels. The MCS defines which modulation is going to be used (QPSK, 16-QAM or 64-QAM) and which code rate will be used (for example ½, ¾, etc). F. Inter-Cell Interference Coordination In order to understand the benefit of using Beamforming one must first understand the concept of Inter-Cell Interference Coordination (ICIC). LTE was planned to allow a frequency reuse factor of 1, which means that all adjacent cells will be using the same frequencies and bandwidths. This causes massive inter-cell interference especially to a user that is located in the cell-edge. Figure 1. ICIC and cell edge effects Without considering ICIC, the SINR would drop rapidly in the cell-edge due to an increase of interference as well as a decrease of useful power from the Serving Cell. One of the main techniques of ICIC is to use Fractional Frequency, which offers a separation of spectral resources within a sector which is partitioned in different geographic G. Beamforming Figure 2. Fractional Frequency Reuse Scheme Beamforming is a technique that combines different radio signals in order to simulate the behavior of a large directional antenna. The virtual antenna can then be directed using an additional signal processing, without the need to physically move the antenna. Beamforming is used to lower interference and increase the overall quality of a telecommunication network. With the use of adaptive beamforming, the main lobe of radiation can be pointed virtually to any direction, and therefore the location of the user should be perfectly known by the enb. The main goals of Beamforming are: Coverage area expansion, where we can cover a larger area because of the higher signal level transmitted. Power consumption reduction in the enb, because power is no longer radiated omnidirectionally and therefore lost since the users only occupy a small part of the sector Improvement of the equity of the coverage of the cells, as the cell-edge users will now be able to receive a signal with much greater quality. Improvement in spectral efficiency, higher SINR provides higher CQI and MCS which allows for an improvement of spectral efficiency. In this work Beamforming is used through the calculation of weights that are applied to the different antennas of the antenna array (up to 8). The weights are calculated in order to get the Minimum Mean Square Error (MMSE) and are heavily dependent on the definition of the location of the user through a technique known as Destination of Arrival (DoA) estimation. The location of the user makes the beamformer create a strong lobe of radiation in his direction and the location of the users in other cells allows for the minimization of power transmitted to them, and therefore minimzes the interference caused to them.

4 H. Quality of Service and Experience Quality of Service (QoS) is very important to operators since it provides feedback from the implemented network and allows for continuous optimization. QoS is defined by Key Performance Indicators measured in the network (signal quality, throughput, latency, jitter, packet loss, etc) and the differentiation of levels of service according to traffic (Voice, Data, etc) Quality of Experience (QoE) is given by the perception of quality that a user sees while using a given network. This is evaluated according to usability, accessibility, integrity, security, etc. A mobile network operator should keep an eye on QoS indicators but should always have in mind that good QoS doesn t necessarily mean good QoE, as it is only a matter of perception. The increased investment in QoS improvement usually leads to an improvement of QoE and therefore a reduction of churn and an increase on overall customer satisfaction. I. Key Performance Indicators 1) RSSI RSSI (Received Signal Strength Indicator) is the total power UE observes across the whole band. This includes the main signal and co-channel non-serving cell signal, adjacent channel interference and even the thermal noise within the specified band. This is the power of non-demodulated signal, so UE can measure this power without any synchronization and demodulation. 2) RSRP RSRP (Reference Signal Received Power) is the linear average of reference signal power (in Watts) across the specified bandwidth (in number of REs). This is the most important item UE has to measure for cell selection, reselection and handover. 3) RSRQ RSRQ (Reference Signal Received Quality) is defined as (N x RSRP)/RSSI, where N is the number of RBs over the measurement bandwidth. 4) SINR SINR (Signal to Noise-Plus-Interference Ratio) is the ratio between the average signal power and the average noise power plus interference power. The interference is calculated from the interferent signals transmitted from the neighbor cells. J. Cellular Structure Like in previous generations, in LTE we consider a cellular structure as the base of network planning. Each cell is represented by a hexagon of equal dimensions and has 3 sectors offered by 3 antennas in the center with 120º of angular distance between them. (5) Figure 3. Cellular Structure of the Network III. RESULTS ANALYSIS Using the model presented, simulations were made to evaluate the benefits of using beamforming as an ICIC technique. Firstly a model with no interferents was simulated in order to give an overall idea of how the SINR changes within a cell. Then neighbor interfering cells were introduced and lastly beamforming was used and SINR was evaluated, giving special attention the user located in the cell-edge. A. Model with no interferents In this model no neighbor cells were simulated and therefore no interferents were considered. This caused the SINR to decrease slowly and the benefits of using Beamforming were not very visible. Figure 4. Results of model without interferents B. Model with interferents In another approach, interferents were considered and with the definition of a fixed Inter-Site Distance (ISD), the position of the user was changed and the SINR of this user was recorded. Figure 5. Results of model with interferents

5 Results show that the decay on SINR occurs much sooner in the model without beamforming and that if we use an array of 8 elements (which is more directive), we can achieve significant gains. The level of SINR in the cell-edge user was registered and the benefits of using beamforming started to be analytically visible. scenario the operator uses the 800 MHz frequency with 10 MHz of bandwidth. The results show an impressive benefit on spectral efficiency, on par with the Throughput performance increase using beamforming with an array of 8 antennas. TABLE IV. CELL-EDGE SPECTRAL EFFICIENCY WITH INTEFERENTS TABLE II. CELL-EDGE SINR WITH INTEFERENTS C. Throughput Calculations Using the previously registered level of SINR calculations were made in order to get the estimation on the achievable Downlink throughput in these scenarios. This estimation was derived from different tables using the following methodology: E. ISD Variation n order to validate the results obtained, another simulation was made where the location of the user was fixed in the celledge and the ISD was variable. The results show that with lower ISD, the Beamforming with 8 antennas provides a higher increase in overall performance and that the benefit of using beamforming starts to fade and the SINR tends to decay linearly at the same pace as the model without beamforming for very large ISD values. Figure 6. Methodology to calculate the Throughput The results were as follows: TABLE III. CELL-EDGE THROUGHPUT WITH INTEFERENTS Figure 7. Variation of ISD for the model with interferents These results show an excellent promise of the benefit of using beamforming to reduce interference, especially if we consider that these results are given per stream of information and therefore in case of using MIMO 2x2 we can achieve a significant +54 Mbps increase in cell-edge throughput. D. Spectral Efficiency Calculation With the results presented previously the spectral efficiency of the system was calculated, taking in consideration that in an urban scenario it s likely that the operator uses the 2600 MHz frequency with full 20 MHz of bandwidth, and that in a rural IV. CONCLUSIONS This work s main purpose was to evaluate the usage of beamforming techniques as an alternative to other ICIC techniques, reducing interference and improving signal strength. Simulations were made and provide an understanding of the evolution of the radio coverage of an LTE system, resulting in a comparison of coverage quality (SINR), performance (throughput) and spectral efficiency between a system where beamforming is used and not used. The performance of a system with Beamforming is directly related to the number of antennas installed, being that with 8 antennas, the array is able to create beams with higher directivity and lower scattering losses, therefore increasing the transmitted power to the wanted user. Beamforming also allows for the reduction of interference to other users in other

6 cells and when these two factors are combined, the results are promising. Overall we achieve up to 30% increased throughput for a cell-edge user, resulting in a very high increase of spectral efficiency to this user. This means that the cell will have a higher equity level of coverage, as the users will not be as easily discriminated in terms of their weak coverage throughout the cell. Beamforming can then be used to increase the quality of experience of a user, since he will be able to experience a higher throughput and, by increasing the coverage are of the cells, the operator can also re-plan the installation of new sites, needing less sites to cover the same areas. The operator should evaluate the cost-benefit of increasing the signal processing power needed for beamforming as well as the increasing the number of antennas in each cell. REFERENCES [1] F. Rezaei, A comprehensive analysis of LTE physical layer, Lincoln: University of Nebraska, [2] M. Ricci, Beam-forming and power control in flexible spectrum usage for LTE Advance System, Aalborg: Aalborg University, [3] P. Carreira, Data Rate Performance Gains in UMTS Evolution to LTE at the Cellular Level, Lisboa: IST, [4] T. Gonçalves, Energy efficient solutions based on beamforming for UMTS and LTE, Lisbon: IST, [5] C. Gomes, Comparação do WiMax e LTE a nível da interface rádio, Lisboa: IST, [6] S. Sesia, I. Toufik e M. Baker, LTE - The UMTS Long Term Evolution: From Theory to Practice, Wiley, [7] N. Wei, MIMO Techniques for UTRA Long Term Evolution, Aalborg: Aalborg University, [8] B. Tiwari, Enabling reuse 1 in 4G Networks, [Online]. Available: