Overview of 5G System-Level Simulation

Since R2024a

System-level simulation (SLS) aids in the design, evaluation, and optimization of a 5G radio access network (RAN). This topic presents:

Key 5G SLS Functionalities

The 5G SLS in 5G Toolbox has these key functionalities.

  • Modeling of data channels and reference signals:

    • Physical downlink shared channel (PDSCH) and physical uplink shared channel (PUSCH)

    • Demodulation reference signal (DM-RS), channel state information reference signal (CSI-RS), and sounding reference signal (SRS)

  • Physical (PHY) layer modeling — Abstracted PHY (link to system mapping) and full PHY

  • Duplex modes — Frequency division duplex (FDD) and time division duplex (TDD)

  • Scheduling strategies — Round-robin (RR), proportional fair (PF), best channel quality indicator (CQI), and custom scheduler

  • Different subcarrier spacing (SCS)

  • Open loop power control

  • Interference modeling

  • Application traffic pattern modeling using file transfer protocol (FTP), On-Off, video, voice over internet protocol (VoIP), and full buffer traffic models

  • Node mobility within a cell by using random waypoint mobility model

  • modulation and coding scheme (MCS), rank, and precoder selection based on measurements performed on the CSI-RS and the SRS

  • Link adaptation

  • Fixed MCS

  • Multiple-input multiple-output (MIMO)

    • Uplink and downlink MIMO

    • CSI Type II based multi-user multiple-input multiple-output (MU-MIMO) with inter-user interference

    • Sounding reference signal (SRS)-based downlink single-user MIMO

    • SRS-based MU-MIMO with inter-user interference

  • Hybrid automatic repeat request (HARQ)

  • Medium access control (MAC) logical channel prioritization (LCP)

  • Radio link control (RLC) unacknowledged mode (UM) and acknowledged mode (AM)

  • Channel Modeling — Clustered delay line (CDL) channel, 3GPP TR 38.901 system-level channel, and custom channel

  • Key performance indicators — Throughput, block error rate (BLER), and spectral efficiency

  • Ability to gather statistics at the application (APP) layer, RLC layer, MAC layer, and PHY layer.

  • Simulation logs and run-time visualizations

Modeling Various 5G SLS Aspects

You can model these aspects of 5G SLS using the node functionalities.

Node and Network Modeling

Node and network modeling consists of these aspects.

  • Model radio access technologies and antenna configurations — System-level simulation models new radio (NR) base stations (gNBs) and user equipment (UE) nodes. For information on how to create NR nodes, see the nrUE and nrGNB objects. The NR stack of these nodes encompasses radio link control (RLC), medium access control (MAC), and physical (PHY) layers. For further details regarding the NR protocol stack, see Composition of NR Nodes.

    The NR Cell Performance Evaluation with MIMO example demonstrates the modeling of a 5G NR cell featuring a multiple-input multiple-output (MIMO) antenna configuration. This simulation involves a set of UE nodes connected to a gNB node.

  • Node placement — In a 5G network planning scenario, the placement of gNB node is crucial for various reasons, including coverage, capacity, and performance optimization. System-level simulation of 5G enables you to set the positions of the gNB and UE nodes using the Position property of nrUE and nrGNB objects.

  • Model mobility patterns — In system-level simulation, you can model UE movement within a cell and study its implications on signal strength and overall network performance. For more information about mobility patterns, see addMobility.

  • Model traffic patterns — In system-level simulation, you can model various application traffic patterns such as file transfer protocol (FTP), On-Off, Video, and voice over Internet protocol (VoIP) traffic models. For example, the Generate and Visualize FTP Application Traffic Pattern example demonstrates the creation of a file transfer protocol (FTP) application traffic pattern. This example showcases the sequence of file transfers with fixed file sizes and variable reading times, offering valuable insights into the characteristics of FTP application traffic models. For more information about traffic models, see the networkTrafficOnOff, networkTrafficFTP, networkTrafficVideoConference, and networkTrafficVoIP objects.

    In addition to these traffic models, you can also configure full buffer traffic for a UE node. See the FullBufferTraffic argument of connectUE object function.

  • Simulate the configured scenario — To simulate the configured network scenario, the system-level simulation uses a wireless network simulator. For more information about this simulator, see Wireless Network Simulator.

For an example of how to model, simulate, and evaluate the system-level performance of the 3GPP enhanced mobile broadband (eMBB) indoor hotspot (InH) scenario, see the Evaluate 3GPP Indoor Reference Scenario example.

Network Scalability

Scalability is crucial for 5G networks, as they must support a large number of connected devices while providing high data rates. System-level simulation helps you evaluate the scalability of a network by simulating the behavior of a large number of devices and evaluating the impact on network performance.

The 5G SLS enables you to create multiple NR nodes in a single object call. For more information about creating multiple NR nodes, see the nrUE and nrGNB objects. Using the SLS, you can analyze factors such as throughput, scheduling fairness, BLER, and spectral efficiency under various traffic loads and deployment scenarios.

For example, the NR Interference Modeling with Toroidal Wrap-Around example demonstrates how to model a 19-site cluster containing a total of 57 cells. Each site consists of three collocated gNBs equipped with directional antennas covering 120-degree areas, resulting in 3 sectors per site. This setup provides an opportunity to evaluate network performance in a relatively large-scale deployment scenario.

Resource Scheduling

The role of a scheduler is to efficiently allocate radio resources to UE nodes. The scheduler operates in the MAC layer of the 5G protocol stack. With 5G Toolbox, you can configure an NR scheduler at the gNB node using the configureScheduler object function of the nrGNB object. You can use the Scheduler name-value argument to specify these scheduling strategies.

  • RR scheduler — Provides equal scheduling opportunities to all the UE nodes.

  • Best CQI scheduler — Prioritizes the UE node with the best CQI.

  • PF scheduler — Offers compromise between the round-robin and best CQI schedulers.

  • Custom Scheduler – Allows you to write your custom scheduling logic. To create a custom scheduler, use a subclass inherited from the nrScheduler class. For an example of how to write a custom scheduler, see the Use Custom Scheduler in 5G System-Level Simulation example.

The built-in scheduling strategies RR, best CQI, and PF also perform dynamic link adaptation for both downlink and uplink. To enable link adaptation for downlink, use the LinkAdaptationConfigDL argument of the configureScheduler object function, and for uplink, use the LinkAdaptationConfigUL argument. For a custom scheduler, you can create your own link adaptation algorithm.

Interference Modeling

Interference is a significant concern in wireless networks, and 5G is no exception. System-level simulation enables you to accurately model and analyze interference in 5G networks. By considering factors such as neighboring cells and channel conditions, you can evaluate the interference levels and their impact on the network's performance. This information can help you optimize resource allocation, scheduling algorithms, and interference mitigation techniques.

The NR Intercell Interference Modeling example demonstrates the process of simulating a scenario involving interference across multiple cells, and assessing the effects on network performance resulting from downlink (DL) intercell interference generated by neighboring cells.

The example NR Interference Modeling with Toroidal Wrap-Around demonstrates a 19-site cluster setup with toroidal wrap-around, per ITU-R M.2101-0, using the 3GPP TR 38.901 system-level channel model. The wrap-around technique ensures uniform interference across the edge of the cluster by creating a continuous looped environment.

PHY Fidelity Levels

The system-level simulation enables you to simulate two different fidelity levels of the PHY layer: full PHY and link-to-system mapping-based abstracted PHY. Full PHY involves waveform generation and decoding, while abstracted PHY models link quality and performance to calculate the packet error rate. For more information about full PHY and abstracted PHY, see Composition of NR Nodes.

The NR Cell Performance Evaluation with Physical Layer Integration example illustrates the use of full PHY processing. This example simulates a 5G NR cell, which includes a set of UE nodes connected to a gNB node. You can also run the example with a link-to-system mapping-based abstracted PHY layer for expedited runtime performance. For an example that uses a link-to-system mapping-based abstracted PHY layer, see the NR FDD Scheduling Performance Evaluation example.

MU-MIMO

MU-MIMO is a key technique in 5G networks for improving spectral efficiency and increasing system capacity. System-level simulation enables you to evaluate the performance of MU-MIMO by modeling the antenna arrays, channel characteristics, and user-pairing in the network. For an example of MU-MIMO modeling, see the NR Cell Performance with Downlink MU-MIMO example. This example assess the system performance of a downlink MU-MIMO with a link-to-system mapping-based abstracted PHY. You can use either SRS or CSI-RS-based DL channel measurements to enable MU-MIMO. To set the downlink channel measurement signal as SRS or CSI-RS, use CSIMeasurementSignalDL of the configureScheduler function. For more information about SRS based downlink channel measurements, see SRS-Based Downlink Channel Measurements for TDD System.

Performance Analysis

By comparing the performance of different network configurations, algorithms, and deployment scenarios, you can gain insight into the behavior of a system and make informed decisions regarding network design, optimization, and resource allocation. System-level simulation enables you to:

Factors Affecting Simulation Execution Time

Multicell Network

A multicell network increases the number of links and introduces additional interference, which increases computational demands and simulation time.

Number of UE Nodes per Cell

Increasing the number of user equipment (UE) nodes per cell raises the number of links, necessitating additional channel estimation and measurements. A greater number of UE nodes leads to greater computational demands and can extend simulation time.

Bandwidth and Subcarrier Spacing

Larger bandwidths and smaller subcarriers spacings increase the number of resource blocks (RBs) to manage, thereby increasing simulation complexity and time.

Number of Antennas at gNB & UE

More antennas at the gNB and UE nodes increases the dimension of the channel matrix, thereby increasing simulation time.

Max Users per TTI (MaxNumUsersPerTTI) Configuration in gNB Scheduler

Increasing the number of users per transmission time interval (TTI) generates more packets, as each packet must be processed for desired and possibly interfering signals. Consequently, the processing time increases for each TTI.

Enabling MU-MIMO Transmission

When you enable CSI-RS-based MU-MIMO using the configureScheduler function, the system automatically uses CSI Type II codebook. CSI Type II codebook offers a significantly higher number of precoders to select from, which increases simulation time. By default, the system uses CSI Type I codebook for single-user (SU)-MIMO.

CSI-RS Report Periodicity (CSIReportPeriodicity) and SRS Transmission Periodicity (SRSPeriodicityUE)

Channel state information-reference signal (CSI-RS) reporting periodicity determines how often a UE node reports channel measurements. In SLS, the UE estimates the downlink channel from the reference signal when the CSI-RS transmission is closest to the reporting time. Users can configure the CSI-RS report periodicity, while the transmission periodicity remains fixed at 10 ms. A low value of CSI-RS report periodicity increases the simulation time because the channel measurement process at the UE node is more frequent. Similarly, a low value of sounding reference signal (SRS) transmission periodicity also increases the simulation time. This is due to the increased frequency of channel measurements required at the gNB node.

Choice of Channel Models

The channel model you choose affects how long simulations take. The free-space path loss model runs the fastest, clustered delay line (CDL) models are slower, and 3GPP TR 38.901 model takes the longest.

Enabling Small-Scale Fading on Interfering Signals

When you configure a channel, such as a CDL or TR 38.901 channel, the signal power computation takes into account both large-scale and small-scale fading. By default, for the interfering links, the configuration considers only large-scale fading. However, when you enable small-scale fading for these interfering links, the simulation time increases. For an example of how to enable or disable small-scale fading, see NR Intercell Interference Modeling.

PHY Fidelity

The choice of PHY fidelity impacts the runtime of system level simulation. Enabling full PHY processing (Through the PHYAbstractionMethod argument of nrGNB and the PHYAbstractionMethod argument of nrUE) increases the simulation execution time. By default, the nrGNB and nrUE nodes use the link-to-system mapping-based abstracted PHY.

Choice of Traffic Models

Based on the data rate requirements, the choice between full buffer and non-full buffer traffic can influence the SLS simulation time. For instance, enabling full buffer traffic to generate high data rate traffic typically results in shorter simulation times compared to using non-full buffer traffic for similar high data rate scenarios (Non-full buffer traffic: networkTrafficOnOff, networkTrafficVoIP, networkTrafficVideoConference, and networkTrafficFTP).

Adding Mobility to UE Nodes

Adding mobility to UE nodes (see addMobility function) increases simulation time. This is because the channel model responds to the dynamic changes in position or velocity. When you do not add mobility to UE nodes, the simulation runs faster.

Disabling Traces

Logging information such as grants, protocol layer statistics, and channel quality for each slot increases memory usage and extends simulation time. Disabling these traces can reduce simulation time by minimizing data logging and associated processing overhead.

Disabling Visualization

Updating visualization options like the resource grid and channel quality for every frame can increase simulation time. Disabling these visualizations reduces the computational load, thereby decreasing simulation time.

PCAP Logging and IQ Sample Capturing

Logging medium access control (MAC) protocol data units (PDUs) into a packet capture (PCAP) file for each packet transmission increases simulation time, particularly as the number of nodes increases. Similarly, periodically capturing In-phase/quadrature (IQ) samples from nodes adds to the computational load, further extending simulation time. For an example of how to enable or disable the packet logging and IQ sample capturing, see the NR FDD Scheduling Performance Evaluation and NR Cell Performance Evaluation with Physical Layer Integration examples, respectively.

See Also

Objects

Classes

Topics