Simulating Mobile Networks Tools and Models Joachim Sachs
Outline Types of Mobile Networks Performance Studies and Required Simulation Models Radio Link Performance Radio Network Performance Radio Protocol Performance Methodology for Mobile Networks Simulation Separation of system properties Flexible and scalable model design Requirements for Simulation Tools Summary Ericsson AB 2005 ITG FG 5.2.1, Mittweida 2
Outline Types of Mobile Networks Performance Studies and Required Simulation Models Radio Link Performance Radio Network Performance Radio Protocol Performance Methodology for Mobile Networks Simulation Separation of system properties Flexible and scalable model design Requirements for Simulation Tools Summary Ericsson AB 2005 ITG FG 5.2.1, Mittweida 3
Types of Mobile Networks Performance Studies Radio Link Performance Mainly radio physical layer Radio Network Performance Radio resource management End-user / application performance Network Capacity and Network Planning (Radio) Protocol Performance Protocol Performance Protocol interactions End-user / application performance Ericsson AB 2005 ITG FG 5.2.1, Mittweida 4
Radio Link Performance (1) Objective Basic performance of the radio transmission multiple-access, modulation, coding Spectral efficiency Error rate Link throughput Algorithms Modulation Channel coding Receiver algorithms Channel estimation (Pilots) MIMO, multi-antenna Ericsson AB 2005 ITG FG 5.2.1, Mittweida 5
Radio Link Performance (2) Models Channel model (Typical urban, suburban, rural) Radio propagation, shadowing, multi-path fading (delay spread) User velocity (Doppler spread) (De-)Modulator Channel Codecs Operates on bit/signal level Single link performance Ericsson AB 2005 ITG FG 5.2.1, Mittweida 6
Radio Link Performance (3) Provides radio link performance models for other simulation studies! Link Performance depending on received signal-to-interference/noise ratio 1 0.1 Indoor A, 64 kbps, 3 km/h -3-2 -1 0 1 2 3 4 5 6 7 BLER with Power Control 0.01 BLER 0.001 0.0001 Eb/No Ericsson AB 2005 ITG FG 5.2.1, Mittweida 7
Radio Network Performance (1) Objective Performance of a radio resource management algorithms System capacity evaluation and network planning Algorithms Admission Control Congestion Control Channel preemption / reconfiguration Power/rate limitations Handover Algorithms Threshold levels for cell handover Macro-diversity (soft handover) Power Control and Link Adaptation Algorithms Adapt power to instantaneous channel condition Adapt coding/modulation to instantaneous channel condition Channel Allocation and Scheduling Resource Reservation Dynamic Resource Assignment Network planning Location of Base Stations Ericsson AB 2005 ITG FG 5.2.1, Mittweida 8
Radio Network Performance (2) Models Channel conditions pathloss, shadowing, multi-path Multi-user interference Multi-cell layout intra-cell interference inter-cell interference User Mobility (Simplified) Traffic Models (Simplified) Protocol Models Operates on ~ms level Largely matrix operations Ericsson AB 2005 ITG FG 5.2.1, Mittweida 9
Radio Protocol Performance (1) Objective Validation, configuration and performance of protocols Interactions of different protocols End-user / application performance Algorithms Radio Protocols Queue Management and Flow Control Segmentation Retransmissions (ARQ) Signaling Procedures Scheduling Higher Layer Protocols Retransmissions Congestion Control Ericsson AB 2005 ITG FG 5.2.1, Mittweida 10
Radio Protocol Performance (2) Models Traffic Models Application Protocols (e.g. HTTP) Transport Protocols (e.g. TCP) Radio Protocols (e.g. RLC, MAC) (Simplified) Channel models Often single cell Possibly single user Operates on >=ms level Detailed protocol implementation Application TCP IP PDCP RLC MAC MAC-d PHY PHY UDP UE Uu RRM, Signaling, Configuration PHY FP FP FP PDCP RLC MAC-d FP GTP-U UDP IP IP GTP-U AAL2 ATM AAL2 ATM AAL2 ATM AAL2 ATM AAL5 ATM AAL5 ATM Node B Iub DRNC Iur SRNC Iu SGSN GGSN Server UMTS RAN Core Network Internet UDP IP Application TCP IP UDP Ericsson AB 2005 ITG FG 5.2.1, Mittweida 11
Example Radio Protocol Configuration Configure protocol parameters and timers Ericsson AB 2005 ITG FG 5.2.1, Mittweida 12
Example: TCP Performance TCP Fully-Reliable Byte Stream Service TCP Mobile Terminal Mobile Network Internet Server RTT Round Trip Time segment segment segment segment segment segment ack ack ack ack ack ack send window Ericsson AB 2005 ITG FG 5.2.1, Mittweida 13
erformance of TCP congestion control TCP Send Window [kbyte] Fast Recovery (transient) TCP Send Window Timeout Slow Start TCP Seq. No Fast Retransmissions TCP Trace Time [s] Ericsson AB 2005 ITG FG 5.2.1, Mittweida 14
Outline Types of Mobile Networks Performance Studies and Required Simulation Models Radio Link Performance Radio Network Performance Radio Protocol Performance Methodology for Mobile Networks Simulation Separation of system properties Flexible and scalable model design Requirements for Simulation Tools Summary Ericsson AB 2005 ITG FG 5.2.1, Mittweida 15
Methodology for Mobile Networks Simulation Approach 1: Separation of interacting system aspects Complementary models and simulators for different kind of studies One simulator can provide simplified models for another one Radio Link Radio Network Radio Protocols Simplified Model Ericsson AB 2005 ITG FG 5.2.1, Mittweida 16
Example 1: Single User Channel Model for WCDMA Power Control Loop 1500 Hz Transmit Power Data Rates from MS Received Power Levels at BTS Pathloss and shadow fading SIR E b /N 0 1 Indoor A, 64 kbps, 3 km/h -3-2 -1 0 1 2 3 4 5 6 7 0.1 BLER with Power Control 0.01 BLER 0.001 0.0001 Eb/No Ericsson AB 2005 ITG FG 5.2.1, Mittweida 17
Example 1: Single User Channel Model for WCDMA TX power RX power Without power control t t With power control t t BLER with Power Control 1 0.1 0.01 0.001 0.0001 Indoor A, 64 kbps, 3 km/h -3-2 -1 0 1 2 3 4 5 6 7 BLER Eb/No Assumption: Perfect Power Control Model: Constant BLER model Limitations: only valid below capacity limit only valid for dedicated channels Ericsson AB 2005 ITG FG 5.2.1, Mittweida 18
Example 2: Capacity-Quality Trade-off for packet data What is Performance of a radio network Quality Coverage vs. capacity Capacity Model requires Realistic traffic model Higher layer protocols (e.g. TCP) for elastic traffic Radio resource management algorithms (admission and congestion control) Inter-cell / intra-cell interference Ericsson AB 2005 ITG FG 5.2.1, Mittweida 19
Example 2: Capacity-Quality Trade-off for packet data Approach 1 Radio Network model inter-cell interference Radio Protocols add RRM algorithms add multiple users for intra-cell interference Approach 2 Radio Network model TCP behaviour Radio Protocols Ericsson AB 2005 ITG FG 5.2.1, Mittweida 20
Methodology for Mobile Networks Simulation Problems: Separation of interacting system aspects Many studies require both radio network and protocol aspects e.g. multi-user scheduling in HSDPA, HSUPA or OFDMA Complementary simulators tend to become increasingly redundant Scheduled user good channel = Radio Link User 1 User 2 #1 #2 #1 #2 #1 #2 #1 bad channel = lo Time Radio Network (Radio) Protocols Ericsson AB 2005 ITG FG 5.2.1, Mittweida 21
Methodology for Mobile Networks Simulation Approach 2: Integrated Simulator with scalable models Single simulator comprises all radio network and radio protocol aspects. Alternative models with different abstractions allow flexible and scalable variety of studies Radio Link Radio Network and Protocol Models of different abstraction level Ericsson AB 2005 ITG FG 5.2.1, Mittweida 22
Simulator Evolution Simulation tool for a new mobile system generally evolves At the start many dedicated (quick) studies of limited scope System specification very unstable, different concepts Different simulators (tools), different models Later larger studies General system specification becomes stable Focus on configuration of system and interactions of different functions Larger simulators, coordinated studies Late stage Full system performance and validation of measurements Full-scale system simulator desirable Multiple simulators often part of the system history Ericsson AB 2005 ITG FG 5.2.1, Mittweida 23
Outline Types of Mobile Networks Performance Studies and Required Simulation Models Radio Link Performance Radio Network Performance Radio Protocol Performance Methodology for Mobile Networks Simulation Separation of system properties Flexible and scalable model design Requirements for Simulation Tools Summary Ericsson AB 2005 ITG FG 5.2.1, Mittweida 24
Common Simulation Tools Radio Link Performance COSSAP General Purpose Languages (e.g. C++) Radio Protocol Performance Commercial/public tools (BONeS, ML Designer, Opnet, ) ns2 not a real option General Purpose Languages (e.g. C++, Java) Radio Network Performance MATLAB Commercial tools (BONeS, ML Designer, Opnet, ) General Purpose Languages (e.g. C++, Java) Ericsson AB 2005 ITG FG 5.2.1, Mittweida 25
Tool Requirements Correct modeling philosophy Radio protocol simulators must be event-driven Radio network simulators can be event-driven Radio link simulators are typically time-driven Stable tools Even commercial tools are still buggy Multi-user and Multi-site development Multiple developers at different locations Revision Control and Merging Support Modularity, Simplicity and Support Basic simulation functions (logging, post-processing, ) Even complex systems should appear manageable with hidden complexity Many part-time users Ericsson AB 2005 ITG FG 5.2.1, Mittweida 26
Tool Requirements Trust in success of the tool Development of a mobile network simulator is a large investment (man power) and should last for many years Discontinuation of a tool can waste a lot of effort and time Easy acceptance in an organization Desirable to spread an advanced simulation tool within the company (other research labs, development units) Low burden to try & play with the tool e.g. tools with license costs require first the decision to use the tool and second the trial Code reuse Desirable to reuse parts of the simulator code, e.g. for an emulator Ericsson AB 2005 ITG FG 5.2.1, Mittweida 27
Outline Types of Mobile Networks Performance Studies and Required Simulation Models Radio Link Performance Radio Network Performance Radio Protocol Performance Methodology for Mobile Networks Simulation Separation of system properties Flexible and scalable model design Requirements for Simulation Tools Summary Ericsson AB 2005 ITG FG 5.2.1, Mittweida 28
Summary Different types of Mobile Network Performance Studies Radio Link, Radio Network and Radio Protocol Performance Different Objectives require Different Models Advanced System Level Studies Require a Combination/Integration of these Models Requirements of Simulation Tools Ericsson AB 2005 ITG FG 5.2.1, Mittweida 29
Ericsson AB 2005 ITG FG 5.2.1, Mittweida 30