Early analysis of UAV using Modeling and Simulation

admin_mirabilis

Early analysis of Unmanned Aerial Vehicle System (UAV) using Modeling and Simulation

Abstract

Architecture exploration, power and performance analysis plays a very important role in the Unmanned Aerial Vehicle system design flow. About 70% of the project success rate relies completely on the architecture that designers select during first 20% of the project schedule and the configurations they work with. The rest depends on the software accuracy and correctness of the algorithms. Architecture can be selected using various approaches such as mathematical computations, formal methods, modeling and simulation. Traditional methods will be useful only if the new design is a small increment over the existing design. Modeling and simulation is a well accepted approach to select system architecture specifically when the system needs to be designed completely a new or when it carries major changes to the original architecture in terms of electronics, network and power characteristics. There is also a need for rapid model development of the proposed system to carry out architecture level analysis studies on power, performance and thermal dissipation. In this document we are introducing VisualSim Architect, a systems engineering software solution from Mirabilis Design for architecture exploration and power, performance, thermal analysis of proposed or existing Unmanned Avionics systems.   

Introduction

The increasing complexity of modern unmanned avionics system architectures in defense requires evaluation of the specification early in the design process. As the lower level of details of subsystem models of the system may not be available at the early stages, a formal method is required to evaluate the performance and platform behavior for the set of input values. In order to achieve this, industry has developed various methods to perform evaluations of given specifications. Modeling and simulation using graphical methods are one of the feasible and effective ways to conduct design space exploration and performance analysis. With a graphical modeling and simulation software platform, designers can visualize complete system architecture and the behavior flow graphically and run interactive simulation runs by varying a number of parameters on the fly. This provides a greater confidence in the system architecture and also acts as a greater level of communication medium.

Simulation Technology

VisualSim Architect is a heterogeneous modeling, simulation and analysis environment. Using this graphical modeling and simulation environment one can develop simulation models of their proposed or existing systems/sub-systems using pre-built, parametrized configurable library blocks. Library blocks include standard set of components used in UAV such as processors, memories, bus, sensor modules, communication channels, backplane, display, data link and etc. Library blocks are provided at both stochastic and cycle accurate level and each libraries are embedded with timing and power information.

When a simulation is completed, the software generates a series of graphical and textual reports that help in activities across different nodes of communication bus, bandwidth achieved, latency across the network and power consumption. This has been deployed at various companies to architect the optimized system that meets user constraints.

VisualSim has more than 2000 report generation abilities including utilization, latency, throughput, bandwidth, thermal dissipation and power. There are also various debugging facilities that are tightly integrated to the software through which user can provide break points, trace activities of devices or network nodes and animate the execution flow.

Systems engineering process

ystems engineering process for conducting architecture exploration and power, performance analysis of an unmanned aerial vehicle using VisualSim can be broken down into multiple subclasses. Initially, the user will create VisualSim models of subsystems and system-wide physical attributes including the power system, thermal aspects and avionics. Next step would be conducting system analysis and exercising range of systems capability and physical attributes as well as variations in behavioral mapping and test cases. Simulation runs determine the compliance of system configurations for top level requirements in terms of power consumption, thermal dissipation, maximum permissible end-to-end latency for real time and hard real time tasks and optimal hardware footprint. VisualSim provides the ability to use the same model to interface hardware in the loop that provides a greater level of system verification. 

One of our recent projects was to conduct domain level analysis of an UAV considering all major subsystems. The requirements can be classified as four parts; Range, Endurance, Maximum Altitude, ISR video quality. As the operating conditions of UAV includes typically four levels, Ascend, Cruise, Descend and Loiter, there is also a need to model these operating modes too, as each levels has different requirements in terms of performance and power. Figure 6 shows the flow diagram of VisualSim model developed to conduct analysis of various subsystems in an UAV.

In addition to modeling of the UAV system to explore its performance, power, functional and thermal characteristics one can also model fault scenarios and random events that represent emergency situations such as enemy aircraft attack, weather conditions and system failure. This enables the user to model and simulate a complete system along with environmental challenges and on-field scenarios.

Figure 6

The domain level analysis is conducted to make sure that each subsystem are meets the expectations and are free from errors and bottlenecks. Let us look at each domains/subsystems and analysis conducted.

Power Generator:

In this module we were interested in capturing output power details and the main constraint was on maximum temperature it can reach. This provides guidelines to select appropriate thermal dissipaters.

Propulsion:

In this module we considered RPM and the maximum temperature it could reach.

Avionics:

This module was quite detailed with all the processor modules, communication modules, sensors and feedback system. Major purpose of modeling this subsystem is to ensure that timing-critical tasks are being executed in the given timing deadline and to identify rejected tasks as shown in figure 7 below. The analysis also helped us in understanding the bandwidth available on the backplane channel with worst-case scenarios. The major components included in this module are Sensor Module, Processor Cord, 1553B, AFDX, CAN and RTOS ARINC 653 with few other resources.

Figure 7: Rejected tasks that exceeds timing deadlines

C&C Data Link:

Command and Control data link module was modeled to identify the C&C range, retries and total power consumption

Video Data Link:

This module was modeled to identify the datalink range for video along with RF bandwidth computation. Here we can also consider having image processing requirement before transmitting the data to the ground system.

System Description

A typical prerequisite to guarantee the safety of flight is a real-time reaction to hazardous conditions. This leads to a number of safety-critical requirements to avionics subsystems that appear like a limitation on the time between a moment when a particular sensor reads actual data and a moment when software processes that data and takes appropriate actions, e.g. delivering a control command to an actuator or informing a pilot. If inability to satisfy such requirement is found during integration tests alone, it may have catastrophic consequences for the project. For example, if an unexpected latency appears on an overloaded bus, it may require redesigning data flows between components and to rescheduling software partitions that lead to redoing significant part of safety analysis and other verification activities, i.e., to significant delays and cost increase for the project.

Model-driven system engineering is considered to be the most promising approach that can help address this concern. The main idea behind it is to represent requirements and architecture decisions in the form of models, i.e., in more or less machine-readable form, and then to apply automated verification techniques or even to automatically derive some parts of the implementation from these models. The benefits are manifold. First of all, automated analysis and verification help to identify multiple problems as early as possible. But even if changes in system design are introduced, models allow for automated impact analysis and reduce effort required for repeated verification.

As avionics plays a very crucial role in UAV functionality we can classify it as three groups of functions, such as;

  1. Flight Control/ Navigation
  2. Flight planning/re-planning for autonomous requirements
  3. Payload management

Flight Control/Navigation itself has a variety of requirements for periodic functions, simple and small computations, use of real time processors and operating systems. For flight planning/re-planning the system may consists of on demand task requirements, very complex and long computations for accurate position and velocity computations. Payload management may not be a complex requirement and can be performed with simple methods.

Modeling and simulation of all 3 major groups of functions will ensure that the architecture is free from critical challenges and prospective bottlenecks. If the design is quite simple with a dedicated resource for each function then in many scenarios computations using spread sheet and other formal methods would be enough. The challenge is really high when multiple tasks are trying to access same resource. This introduces contention for resources and directly impacts on the performance if tasks are not distributed correctly.

Solution Description

Purpose of modeling and simulation of a highly complex system such as a UAV is mainly due to one major factor, Resource Limitation. Resource limitations can be explained typically in terms of Processing speed, communication bandwidth, contention for processing and communication resources, storage and buffer access times, power and thermal factors. These factors result in high system dependency and causes system unpredictability, thereby impacting the system performance and power requirements directly.  Modeling and simulation of both hardware architecture and software behavior flow early in the design flow ensures that the system is optimal enough to process real-time and hard real-time tasks  handling timing-critical applications. VisualSim provides a framework for modeling both Hardware architecture that consists of processor card, graphics cards, IO cards, communication channels and even sensor modules on a UAV. Using file based or flow diagram based approaches, the user can map their application onto the target hardware platform and run simulations to conduct power, performance and functional analysis even before the software and hardware is ready. The customizable and parameterizable modeling framework integrates both standard and custom hardware resources thereby providing complete flexibility for considering custom hardware devices during early system modeling. 

A sample hierarchical view of UAV model developed for conducting analysis on 1553B is shown in the figure below.

Figure 1: UAV model for analyzing 1553B

Resource details of the Air-Link within the UAV include RTOS, Cache, Processor Array and Disk. The number of processors is a parameter that can be varied. The model handles the contention and arbitration for resources. In addition, each frame has a priority that impacts scheduling at the RTOS and at the Processor Array.

The sample model shown in figure 1 has two onboard datalinks and is connected to 1553B remote terminals. The data is sent across a noisy channel to the terrestrial system. Internal view of the Datalink subsystem is shown in figure 2.

Figure 2: Sensor modules and Datalink

Each aircraft sensor is modeled as a uniform distribution traffic generator producing packets within a window for each event ranging from a minimum time set by parameter to a maximum time equal to twice that of the minimum time (and adjustable parametrically to any other desired discrete value or ratio). Each sensor also stores the sensor data locally and can retransmit, if requested by the Ground Station, or if an acknowledgement of reception is not received within a parametrically specified time interval.

The Terrestrial System tests for unique arrival of each transmission, checksum for transmission errors, compute latency and acknowledgement for retransmission. The Terrestrial System uses transmitted distance from UAV to the ground vehicles to determine if the frame must be read from the regenerated satellite version or directly from the aerial vehicle.

The Simulation result provides details on 1553 B throughputs, latency across 1553B, Sensor to Terrestrial System latency and processor utilization. Figure 3 shows reports on 1553B throughput, Sensor to Terrestrial System latency.

Figure 3: Latency and throughput reports

Second part to the system level model development and performance analysis would be computation of power consumption and thermal dissipation. This provides a great visibility to the power factors of the system and enables users to make sure that the system requirements are met. Sample VisualSim model constructed for conducting thermal analysis is shown in figure 4.

Figure 4: VisualSim model of UAV for Performance and Power analysis

Simulation run provides varieties of statistics and graphical reports that allow designers to make required modifications to the architecture to meet requirements. Sample reports are shown in figure 5.

Figure 5: Latency and details on power consumption by subsystems

Simulation of enemy aircraft tracking, processing and decision making can be modeled to identify the possible bottlenecks that the proposed system may pose. Figure 6 shows the sample internal structure of an inertial sensing system consisting of sensor and control environment, gyro interfaces, gimbal motors and accelerometers. Figure 7 represents a graph that shows tracking of enemy aircraft and Figure 8 represents the processing carried out in the system to address the enemy aircraft threat, here it  represents once the enemy aircraft approaches the UAV deviates from its current path.


Figure 6: Inertial Sensing System
Figure 7 & Figure 8

The simulation results also provided various other statistics including the utilization of resources on board, latency and bandwidth available on backplane. Table 1 represents the reports with different system configurations.

Table 1: Comparison between simulation runs

Conclusion

Complex systems reflecting competitive products have been found to be highly unpredictable.  To cope with this unpredictability, customers have successfully employed system-level design and analysis techniques to the optimal development of these systems using VisualSim Architect.  The use of VisualSim Architect in conducting system-level design represents a proven industrial solution that enables rapid convergence to an optimal, and therefore, highly competitive product.

Trade-off studies of this kind can help in the analysis of the system performance for a variety of operating conditions. Early understanding of optimal performance, power consumption and thermal dissipation of each sub-system can save significant time in prototype testing and deliver much more robust system operation. This methodology allows the system engineer to start with an abstract concept and increase model fidelity and accuracy through successive levels of abstraction. Establishing a simulation platform allows for quick and accurate trade studies and spiral engineering enhancements over the program’s lifecycle. This also allows investigations into interoperability which reduces total overall cost of the electronic systems. This is especially important for electronics where different system elements and components face obsolescence or need replacement for enhanced performance.