Performance review of Pico base station in Indoor Environments

Transcription

1 Aalto University School of Electrical Engineering Performance review of Pico base station in Indoor Environments Inam Ullah, Edward Mutafungwa, Professor Jyri Hämäläinen

2 Outline Motivation Simulator Development Winprop Tool Simulation Parameters Performance Metrics Simulation Results Conclusions 2

3 Motivation Traditional Mobile network s coverage and capacity are insufficient for high data services especially in indoor environments. The HetNets (Heterogeneous networks) is one of the solution by adding small cells to bring network closer to user and enhance the system performance. Pico base station deployment can ensures good quality of in-building coverage & capacity.

4 Why Winprop tool? Credible channel models are difficult to find for lower frequencies as well as for street level interference propagation. WinProp is standard software tool in the domain of wireless propagation and radio network planning for different environments. Includes different channel models to predict the path losses for geographic areas.

5 Simulated Area (Winprop Tool) Four tri-sectored macro base station (MBS) Building of interest comprises five storeys with several buildings around Pico base stations (PBS) deployed inside the building

6 Pico Deployments Location 1 Location 2

7 Winprop Simulation Parameters PARAMETER VALUES/ MODELLING ASSUMPTIONS Simulation resolution Indoor areas: 1 m Outdoor areas: 10 m Height of prediction results Building properities Pico base station parameters Operating frequency (for both macro and pico) Indoor areas: 1.5 m above floor level Outdoor areas: 1.5 m Materials include concrete, wood, steel, brick etc. Individual material properties (permeability, vertical/horizental transmission losses, diffraction losses etc.) defined in WinProp (WallMan) Deployment height: 1.5 m above floor level Directional antenna pattern (Weighted Bilinear Interpolation (WBI)) Transmision power: 2 W Antenna gain: 5 dbi Antenna azimuth: Pico 1 (90 ), Pico 2 (270 ) Antenna downtilt: 0 for all picos 800 MHz & 2100 MHz Propagation modelling Dominant path model used for both indoor and outdoor areas Waveguiding effect considered (for distance upto 20 m)

8 Winprop Simulation Parameters Access Point Antenna Heights (m) Macro Site 1 Antenna 1 50 Macro Site 1 Antenna 2 40 Macro Site 1 Antenna 3 40 Macro Site 2Antenna 1 25 Macro Site 2 Antenna 2 25 Macro Site 2 Antenna 3 25 Macro Site 3 Antenna 1 25 Macro Site 3 Antenna 2 25 Macro Site 3 Antenna 3 25 Macro Site 4 Antenna 1 50 Macro Site 4 Antenna 2 50 Macro Site 4 Antenna 3 50 Pico Base Station (First floor) 2.5 Pico Base Station (Second floor) 6.2

9 Matlab Simulation Parameters Bandwidth Parameter Value 10 MHz,48 PRBs for Data & 2 PRBs for signalling Thermal Noise Power Spectral Density -174 dbm/hz SINR efficiency (SINR eff ) 0.85 Bandwidth efficiency (W eff ) 0.42 Antenna Configuration MIMO 2*2 MBS Parameters MBS Transmit Power 46 dbm Pico Base Station Parameters Pico Transmit Power 33 dbm

10 Simulated Maps This slide shows the received power maps of first floor for PBSs. Both picos are deployed at the first floor at location 1 & 2. In the given maps, it is noticed that PBS have good received signal level in the part of buildings near around the PBS. Pico 1 (1st floor) Pico 2 (1st Floor)

11 Simulated Maps This slide shows the received power maps of second and fifth floor for pico 1 base station. In the given maps, it is observed that the received signal power level decreases with the increased height of the building, due to multiple reflection of signal with inner structure of the building. Pico 1 (2nd floor) Pico 1 (5th Floor)

12 System Methodology In current study, two floors (Floor 1 & 2) has been assumed, where 4 user equipments (UE) has been dropped randomly in each of two floors. All indoor UEs are force to connect to Pico even in parts of the building where the Macro signal might be stronger. Though they may connect to PBS with good reference signal received power (RSRP). Round robin scheduling is used to assign network resources to each UE. Two different scenarios proposed for PBS deployments. Single Pico deployment (i.e.; Pico 1 at location 1, floor 1) Two Picos deployment (i.e.; Pico 1 at location 1, floor 1 & Pico 2 at location 2, floor 2)

13 Performance Metrics The performance metrics used in this study are the cumulative distribution function (CDF) of; SINR (Signal-to-Interference- & Noise-Ratio) per PRB Indoor User Equipment (UE) Throughput Simulation has been done for two carrier frequencies. (i.e. 800 & 2100 MHz) in two difference interference environments. When Indoor Pico UEs receive Co-channel interference from MBS. When Indoor Pico UEs don't receive the Co-channel interference from MBS.

14 Single PICO deployed at Location 1,Floor 1

15 Single Pico deployment at Location 1, Floor 1 Pico 1 at Location 1, Floor 1

16 CDF of SINR per PRB & UE Throughput (800MHz) This slide shows the SINR and throughput performance of indoor UEs in different interference environment at carrier frequency of 800 MHz. Comparing blue and black curves, It is observe that PBS yields worse SINR and throughput performance upto 70 % indoor UEs due to reception of high co-channel interference from MBS (Its worthy to mention that 4 indoor UEs randomly deployed in second floor.). Though, PBS provides good coverage in first floor. Moreover, the PBS have significantly improved the indoor UEs performance in terms of SINR as well as throughput, if orthogonal operating frequency resources are allocated for PBS. Here the Picos UEs do not receive co-channel interference from MBS.

17 CDF of SINR per PRB & UE Throughput (2100MHz) This slide shows the SINR and throughput performance of indoor UEs in different interference environment at carrier frequency of 2100 MHz. It has been noticed that PBS and MBS have same performance levels as discussed in previous slide for 800 MHz.

18 Two PICOs deployment Pico 1 at Location 1, Floor 1 Pico 2 at Location 2, Floor 2

19 Two Pico Deployments Pico 1 at Location 1, Floor 1 Pico 2 at Location 2, Floor 2

20 CDF of SINR per PRB & UE Throughput (800MHz) This slide shows the SINR and throughput performance of indoor UEs for the case, when PBS are deployed in both floors at proposed locations (i.e., location 1 & 2). Given results show that system performance has been improved by deploying two PBS as compared to Macro only networks. Moreover, the performance gain can be improved by operating the PBS on separate carrier frequencies rather then sharing frequencies with MBS. Though, the UEs served by each PBS receive interference from each PBS. Pico interference included Pico interference included

21 CDF of SINR per PRB & UE Throughput (2100MHz) This slide shows the SINR and throughput performance of indoor UEs in different interference environment at carrier frequency of 2100 MHz. Comparatively, It has been noticed that PBS and MBS have same performance levels as discussed in previous slide for 800 MHz. Pico interference included Pico interference included

22 Single Pico versus Two Picos(800MHz) In this slide, the comparative analyses of single versus two PBS deployment was done. The PBS has been allocated separate operating frequencies then MBS. Hence, the indoor PICO UEs only receive co-channel interference from the non serving PBS (in case of two PBS deployment). Results show that comparatively single Pico enhanced the system gain in terms of SINR as well as UE throughputs (on average level ) then two Picos deployment. The reason is that indoor Picos UEs still receiving co-channel interference from the non serving PBS (in two PBS deployment), because, both the PBSs are sharing the same resource pool. Hence, single pico deployment will yields improved gains for the two floor cases. Need to study more. Pico interference included Pico interference included

23 Conclusions System is heavily interference limited in both cases 800 & 2100 MHz. Pico performance gain is small, when its shares network resources with MBS (Due to Co-channel interference from nearby MBS). SINR and UE throughput levels are very large, when orthogonal resources are allocated for PBS (Co-channel interference omitted from nearby MBS). Hints concerning to indoor public safety (PS) relaying If PS system applies its own frequency and macro network deployment is scarce, then indoor relay may provide high data rates on access link. Though, the backhaul link is still bottleneck. Hence, relaying is feasible, if resource split is enabled between direct and access link.

24 Revised Plan for HEWINETS-WP1 Item / Scenario Remarks Status 1 Review of channel models for street level propagation in 400 MHz 2 Comparisons between 400MHz and 2600MHz propagation especially in street level propagation Winprop Ray tracing tool has been used to calculat the path loss estimation and Implemented in Matlab tool for simulations Winprop Ray tracing tool has been used to calculat the path loss estimation and Implemented in Matlab tool for simulations Done Done 3 Indoor penetration comparison for 10 MHz bandwdith at carrier frequencies 400, 800 and 2000 MHz. 4 Indoor penetration comparison for 5 MHz bandwdith at carrier frequencies 400, 800 and 2000 MHz. 5 Local impact of interference investigated using e.g. blocking area around the relay as a measure. 6 Variation of required TX power in relay related to applied carrier frequency 7 Guide: How to read performance results? Comparative results for relaying performance has been generated via Matlab simulations Comparative results for relaying performance has been generated via Matlab simulations Free space path loss model were used for calculating the relaying interference experienced by outdoor commercial UEs located around relay in the radius of 50 meter. Propagation models first needed. Some studies might be possible to do using ray tracing tool. Detailed report has been written on plots & figures, in order to help the reader to understand results. Done Done Done Undone Done 8 Uplink results for single relay case This requires development of uplink relay simulator Undone

25 Thank you very much for your time

26 Supporting Slides

27 Pico Antenna Pattern 1/3 We could start with a simple pattern of the figure below. This is horizontal antenna pattern 5dBi G H (q ) 0dBi -180deg -100deg -70deg -15deg 15deg 70deg 100deg 180deg -10dBi

28 Pico Antenna Pattern 2/3 This is the vertical antenna pattern, denoted as G V (y ) 0dBi -3dBi -180deg -100deg -70deg -15deg 15deg 70deg 100deg 180deg

29 Pico Antenna Pattern 3/3 Total antenna pattern is of the form G ( q, y ) = G ( q ) + G ( y ) H Note that gain pattern is given in decibel scale. Pico transmission power is 2W (=33dBm) V

30 Winporp Antenna Pattern 2D antenna partterns (vertical and horizental) should be converted to 3D pattern in WinProp for propagation modelling Four different conversion algorithms available in WinProp Algorithm 1: Arithmetic Mean (AM) Algorithm 2: Bilinear Interpolation (BI) Algorithm 3: Weighted Bilinear Interpolation (WBI) Algorithm 4: Horizontal Projection Interpolation (HPI) We compared the conversion results of all algorithms and have settled for Algorithm 3: WBI Most accurate conversion, particularly in the horizental plane Further implementation details of algorithm here: F. Gil et al, A 3D Interpolation Method for Base-Station-Antenna Radiation Patterns, IEEE Ant. & Prop. Mag., Vol. 43, No.2, April 2001.

31 Winporp Antenna Pattern 1/3 2x2D pattern to 3D pattern conversion result for using WBI algorithm

32 Winporp Antenna Pattern 2/3 Plot of horizental pattern after conversion 3D pattern conversion

33 Winporp Antenna Pattern 3/3 Plot of vertical pattern after conversion 3D pattern conversion