As communication networks become ubiquitous, there is continuous innovation both in the technologies they employ and the services they are expected to provide, creating new challenges for equipment manufacturers, service providers, network planners, and operators. Network simulation can assist in addressing many of these challenges, including:

• Comprehensively assess network behavior under different operating conditions, such as types and volume of traffic (e.g., voice, video, file transfer) to identify potential problems, understand the root cause(s), and to evaluate potential solutions before deployment.
• Provide a zero-risk, low-cost means to assess the cyber resilience of the network.
• Assist developers of new applications and services by providing a low-cost testbed to evaluate their performance under different conditions (e.g., mobility in dense urban terrain) and to assess their interoperability with competing applications (e.g., potential interference between a multi-player video game and voice-call handoff).

Requirements for Network Simulation Tools

In order for a network simulation tool to add value and provide a high return on investment for users – network operators, planners, designers, testers or analysts – it must satisfy the following requirements:

Accuracy

Network simulations must be able to accurately reproduce the behavior of the corresponding network as it operates in the field. To achieve this, the simulator must model the network devices, end-to-end protocols, traffic, and the operational environment, accurately and at the appropriate fidelity. Simulators that use abstractions to reduce model execution time (e.g., by ignoring terrain effects or buffer size or bursty traffic patterns) will fail to capture the network dynamics which significantly affect network performance and lead to potential issues like congestion, voice quality degradations, or failures.

Speed and Scalability

Multiple simulation runs are typically needed to obtain statistically valid and meaningful results; thus fast execution speed, without sacrificing accuracy, is essential. As network size increases, the simulation time often increases non-linearly with size and traffic volume. As network dynamics of large networks are very different from smaller networks, extrapolation of results obtained by simulating small networks to larger networks is difficult.  Therefore, for the simulation tool to be useful, it must be able to run ‘at-scale’ models faster than real-time without sacrificing accuracy.

Ease of Creating Simulation Models

Developing high-fidelity models of network devices and protocols from scratch is time-consuming and also requires detailed networking knowledge together with programming expertise. Thus, to provide a high return on investment, network simulators must facilitate rapid model creation.

• The simulation tool should provide a library of pre-built, accurate models of a diverse set of network components that are easy to customize, such that the resulting simulation model is an accurate representation of the target network. The model library should include models for:
  • Devices such as routers, servers, access points, base stations, hosts, mobile devices, radios, etc.
  • Common applications, such as web browsing, email, messaging, database transactions, file transfer, video conferencing, VoIP, etc.
  • Wired networks and wireless communication technologies such as Wi-Fi, WiMax, cellular (GSM, UMTS, LTE, 5G), satellites, etc.
  • Cyber attacks and defenses, and host vulnerabilities which can be exploited by cyber threats
  • Protocols at all layers of the protocol stack, including routing protocols used in different domains, protocols for QoS provisioning, MAC protocols, etc.
  • Environmental factors which affect wireless communications, such as terrain and weather effects, and mobility patterns of mobile devices
• The network simulator must provide an intuitive GUI to rapidly create complete scenarios from the library of network components.
• It must be easy to modify operational parameters of a ‘baseline’ scenario and create different ‘what-if’ scenarios.
• To analyze existing or ‘as-is’ networks, it should be easy to develop simulation models from representations of the networks in popular, standard formats, such as Visio files. This will ensure that the simulation model accurately represents the network being analyzed.
• To enable accurate modeling of network traffic, the simulation tool should support import of representations or traces of traffic from operational networks and faithfully reproduce it in the simulation model, for example, by importing PCAP files generated by Wireshark or other similar tools.

Ease of Network Visualization and Performance Analysis

The results from a network simulation are used for network analysis and to provide insight into network operations. To facilitate this, the simulation tool should have the following features:

• A database of comprehensive and detailed statistics which can be used after the simulation to generate actionable reports. These are essential for identifying issues, understanding the root cause, and to compare alternative mitigations.
• An operational visualization capability to provide an overview of network operations by displaying end-to-end packet transfers and key statistics which are updated dynamically as the simulation is running. Such visualization can help understand the interactions between different applications.
• A technical visualization interface which provides details of network operations by graphically displaying network performance metrics like coverage heat maps, the paths traversed by packets, states of queues, etc. Such visualization can be very useful to identify and resolve performance issues, such as network congestion.
• Interface to third party analysis, data visualization, and report generation tools, such as Tableau

Ability to Interface with Live Applications

The network simulation must be able to interface with live network components and real applications. This in turn requires that the simulation run in hard real-time.  In other words, the application will interact with the simulated network exactly as it interacts with the corresponding live network.  This enables an analyst to directly assess the application’s performance under different operating conditions, such as different terrains or mobility paths, simply by modifying appropriate parameters of the network model. Similarly, such simulations can be used to investigate potential interference among applications under different operating conditions like varying levels of traffic intensity. This provides a low-cost and high return on investment alternative to performing such assessments using physical testbeds.

SCALABLE’s network simulators, QualNet and EXata, can run accurate, at-scale network simulations faster than real-time, support an extensive library of network models, and provide a suite of tools for scenario creation and network visualization and analysis.