Audio Visual Capture And Distribution For Experimentation And Analysis

Mr. Simon Crase

Land Operations Division, Defence Science and Technology Organisation simon.crase@dsto.defence.gov.au

Abstract. Experimentation and Analysis, utilising modelling and simulation, forms a core component of the

work performed within Land Operations Division at the Defence Science and Technology Organisation (DSTO). Much of this work is performed within the Army Experimental Framework (AEF) program which is currently exploring army structures and capability for the Army Objective Force which will come into being circa 2025.

Audio visual capture and distribution is a tool that has long been utilised within DSTO experimentation to allow analysts to review why key decisions were made to assist with their analysis. These recordings are particularly useful in post analysis of both the mission planning phase of Army experiments, and the execution of the commander’s intent during the wargame or simulation.

Over the last 10 years, substantial progress has been made in the technologies, tools and techniques used for audio visual capture within Land Operations Division at DSTO, and this paper outlines this progress and some of the lessons learned. This included the development of a Mobile Audio Visual Capture Suite (MAVCapS), the transition from VHS tapes to DVDs to current computer based recording technologies that allow live streaming over computer networks, easy management of the recordings, and easy access to the recordings by the analysts.

Also discussed are developments in camera and microphone technologies and the transition from consumer camcorders, to fixed CCTV cameras, to remotely controlled ‘Pan Zoom Tilt’ cameras, and the challenges of using IP based cameras for video recordings.

1. INTRODUCTION Audio and video capture and distribution is a useful tool for supporting simulation and military experimentation. Land Operations Division (LOD) at DSTO have been using audio visual capture and distribution for over a decade and this paper details some of the advantages this provides and some of the lessons learned over the period.

Within LOD, audio visual capture has primarily been utilised by the Army Experimental Framework task. This task performs regular wargames and simulations in support of analysis assessing future army capabilities. The experiment vignettes typically consist of military members performing mission planning, briefing the plan of action, playing out the wargame using an appropriate simulation, and then an after action review (AAR). Audio visual capture and dissemination can assist the DSTO analysts through all phases of the experiment.

1.1 The Value of Audio Visual Capture The audio visual capture or recordings can assist the analyst by allowing them to review the video recordings of the experiment when performing post experiment analysis and producing reports. It allows analysts to go back and review why decisions were made by the military participants. This information can be invaluable when unexpected result come out of the wargame simulation.

Video capture is also a key tool for human factors orientated studies. Analysis of human behaviour or interactions within a simulation can be difficult to document and capture in a live environment. Having a capability for analysts to review actions, potentially in slow motion, or allowing pauses for documenting observations is a valuable enabler for capturing accurate results for further analysis.

Audio visual recordings can also be used for the general documentation of experiment proceedings which may be required for record keeping or auditing purposes. This footage could also be of value for promotional or marketing purposes, subject to release approval.

1.2 The Value of Audio Visual Distribution Besides the audio visual recordings, live distribution (or streaming) of the video can also be of significant value to experiments and simulations. Audio video distribution is the transmission of live video to other locations. Within LOD experiments, these are normally other rooms within the simulation facility.

Distribution is valuable for several purposes. It can provide situational awareness to the experiment coordinator and facilities managers. It allows analysts to observe experimental proceedings without interfering with the experiment through their presence. This can be of particular value for human factors analysts observing participants actions or interactions within a simulation. The presence of an analyst (or other influential people such as senior military members) can often skew the subject’s behaviour so remote observation removes this

influence. There have also been cases where video has been transmitted live to DSTO locations interstate. This has allowed a wider audience to hear briefings or AARs. This is particularly valuable for team members who are involved in only a small part of the experiment so could not justify the time or financial expense of travelling to the experiment.

We have detailed the value in audio visual capture and distribution, and we will now explore the options for implementing this and some of the techniques used by LOD over the years.

2. IMPLEMENTING AUDIO VISUAL SOLUTIONS When considering implementing an audio visual capture or distribution system, the key point to focus on is the requirements of the particular experiment or simulation. The recommended way to understand these requirements is to have a ‘data collection and analysis plan’ and have this integrated with the wider ‘experiment plan’ (1). To produce this data capture and analysis plan, the team will be involved with the experimental planning process and will build an understanding of the required outputs. It is strongly recommended that this planning process start as early as possible. Some of the questions to consider for the data collection team are:

• What is the intended use of the video? • How many locations will need to be recorded, for

how long, and will they occur simultaneously? • How/When would users or analysts need to access

the video? • What staffing and skill set is available to

implement the chosen solution? • What staffing and skill set will be available to

operate the equipment during the experiment?

Once the requirements are understood, development of a solution to meet those needs can begin. All required equipment should be document together with its proposed location, each cable run, and a block diagram laying out the system design (2).

3. LOD SYSTEM IMPLEMENTATIONS TO SUPPORT EXPERIMENTATION LOD has been using audio visual capture to support experimentation for over 10 years. Over this time, we have used a wide variety of solutions that have changed as requirements and technologies have evolved. This section aims to detail that development, the options available, and why certain options were chosen. The locations used for the experiments have been a variety of facilities, primarily in Puckapunyal Army base in Victoria and DSTO Edinburgh in South Australia. Often, the experiments are held in different rooms within the facilities each time or the facilities have been used for other purposes between experiments so the audio visual systems needed to remain flexible and removable and permanent installations have not been

appropriate. The solutions also need to remain scalable and potentially usable in a variety of environments.

3.1 Camcorders During early LOD experiments, camcorders and tripods were used extensively, and occasionally still are. Camcorder simply refers to the common handheld video camera used for home videos. The name camcorder comes from the fact that it combines a camera and recorder in the one device. Camcorders provided a large degree of flexibility in where they could be set up, and they were quick to configure.

When originally used by LOD, Hi8 format tape camcorders were the most common. A large number of recordings were made using these and these tapes are still stored at DSTO. Disadvantages were that these camcorders required a large amount of staffing to regularly change tapes and batteries in the cameras. They were also fairly intrusive, requiring a tripod that took up valuable space in some of the smaller rooms.

An important factor when considering the use of a camcorder is what format and media the camera uses. There are many different recording formats and media used in modern video cameras and it is important to consider the differences and potential benefits of one system over another. Aspects to consider are; how the recordings will be played back; how long can the recordings be; will there be a large number of recordings or media to manage; is the recording media reusable; will the recordings need to be permanently archived, and if so, how will that occur.

While the recording media is one of the major characteristics to consider when selecting a camcorder, there are also several other considerations. One is the power requirements. If extended periods of recording are required, can a power adaptor be connected to the camcorder, or will it need multiple batteries that are changed regularly? Another factor, depending on the requirements, is whether the camera has a live video output capability for distribution to other locations.

Today, there are a variety of camcorders available within LOD to suit a variety of requirements. Camcorders are primarily used for very small scale experiments or for recordings in external experimentation facilities. Their small size allows them to be transported easily in aircraft luggage and quickly set up in new locations. Occasionally, their recordings are later digitized and made available for playback over the DSTO computer network.

3.2 Fixed CCTV Camera Systems Beyond the simple camcorder, there are the large numbers of systems designed for video surveillance. These are commonly referred to as Closed Circuit Television or CCTV systems. These range from simple fixed cameras and recorders designed for home or store security recordings through to fully integrated, centrally controlled systems such as those used in large commercial facilities or government infrastructure.

Fixed cameras refer to the stationary style surveillance cameras that point in one fixed direction. They come in various styles and sizes, and are most commonly analogue cameras that output PAL or NTSC video over a coaxial cable (although digital versions are available). Fixed cameras are typically quite affordable, and a wide range of lens options are available. The wide range of lenses allows specific setups to be developed based on the requirements for the experiment. Potential drawbacks of fixed cameras are that they require extensive planning and setting up (zooming and focusing) for the specific target that requires capturing and they can be inflexible if the experiment setup changes.

One issue with utilizing security orientated systems for capturing experiments and simulations is they often do not contain audio capabilities (2). Systems typically only have cameras but no microphone. If audio capture is required, external microphones may be required. There are a wide range of microphone solutions available so which ones are best suited to the particular audio visual capture requirements need to be assessed. Highly directional ‘shotgun’ microphones can be valuable in recording audio from an individual while cutting out background noise. A lapel microphone may also be suitable, potentially combined with wireless transmitters. For larger groups, omnidirectional ceiling or table located microphones may be more suitable, or even an array of microphones. The approach of having microphones independent from the camera allows the most suited microphone technology to be utilised for each specific situation.

During a series of experiments in 2000 aimed at restructuring the Army, LOD started implementing some solutions using fixed CCTV cameras instead of camcorders. These CCTV systems had various cameras and microphones mounted to the ceilings of the planning and wargaming rooms. The cables ran back to an adjacent observer room where TV monitors and speakers allowed analysts to view live video as well as VCRs recording the proceedings. Video switch boxes allowed the analysts to change views between the separate cameras. While this system suited the needs of the time and meant there was minimal intrusion into the experiment, it was very labour intensive to set up and configure and then pack up after the experiment.

3.3 Mobile Audio Visual Capture Suites (MAVCapS)

To ease the workload in setting up and configuring the audio visual capture and distribution equipment for each experiment, LOD developed a system in 2001 called the Mobile Audio Visual Capture Suite (MAVCapS). The aim of the MAVCapS was to integrate all of the required components of the audio visual capture and distribution system into a single mobile package. As shown in Figure 1, this took the form of a 19” rack housing with wheels added into which all of the required components could be mounted and permanently configured to operate together. Then

the MAVCapS could simply be wheeled into the room where it was required and connected to cameras and microphones to perform the recordings. Fourteen of the MAVCapS were built by the Scientific Engineering Services within DSTO.

Audio PreAmp .

VHS VCR

TD- 100 Time Date Stamp

Audio Mixer Drawer

PZT Controller Drawer

Mobile Audio Visual Capture Suite DSTO MAVCapS #x

TV Monitor

Distribution Amp

IR T

rans

mitt

er

19" Shelf 19" Shelf

19" Shelf 19" Shelf

PZT Controller Multiplex

4 - 1 Video Switch

Figure 1: Initial MAVCapS Configuration The initial components within the MAVCapS were:

• TV Monitor – to view the video • Video Switch – to select which camera to view • Audio Pre-Amplifier – to provide amplification for

microphones • Audio Mixer – for selecting and mixing audio from

multiple microphones or sources • Audio and Video Distribution Amplifier – to

output the audio and video to multiple locations such as the monitor, headphones, the recording device, or for wider distribution.

• Video Cassette Recorder (VCR) for recording the audio and video.

These MAVCapS would be connected to cameras and microphones that were temporarily attached on the ceiling of the experimentation facility. Cables were run across the ceiling and down the wall to the location of the MAVCapS and then recordings could be made.

3.4 MAVCapS Evolution Over time, improvements and adaptations have been made to the MAVCapS and the overall audio visual capabilities.

3.4.1 Introduction of Pan Zoom Tilt Cameras One significant upgrade was the use of Pan Zoom Tilt (PZT) cameras. This involved adding a joystick camera controller to the MAVCapS and one or more PZT cameras to the facilities ceilings. PZT refers to cameras that can be remotely controlled for the direction they face (pan and tilt) and their zoom. This added significant flexibility to the way proceedings could be recorded, meaning action in any part of the room was

recordable. A similar capability using fixed cameras may require a large number of cameras facing many directions. Drawbacks of PZT systems are that they are often significantly more expensive and also more complex. A control station or additional equipment is often required to control the camera, as was integrated into the MAVCapS.

3.4.2 Network Cameras Another camera option available is network cameras (or digital or IP cameras). Network cameras use computer networks to transmit their video instead of the traditional coaxial cable of analogue CCTV video surveillance systems. Network cameras can come in fixed or PZT format. Typically, they connect via Ethernet cable although wireless versions are available. Often, they have an inbuilt web-server that lets users see their displayed video by simply browsing to the cameras web address or IP address. This makes them particularly suitable for simple video distribution. Disadvantages are that the format is not as simple to record using traditional video recording devices and quality can be limited by network bandwidth. Network cameras have been tested at LOD but the difficulty in recording using them with current equipment meant they did not suit our requirements. If developing a new system with no legacy equipment constraints, network cameras may be more suitable and should be considered.

3.4.3 MAVCapS Recording Device Upgrades A significant area for upgrade over the life of the MAVCapS has been the recording devices. This started with the replacement of the VHS recorders with DVD recorders as they came onto the market. This allowed higher quality recordings and playback of the DVDs on analyst’s computers.

Another option for video recording is Digital Video Recorders (DVRs). These hard drive based recorders offer large storage capacity but it can be difficult to get video on and off of the device for reviewing elsewhere or for permanent storage. Hence, they did not suit the requirements for the MAVCapS.

A final option for recording is computer based capture using video capture cards. This offers the advantage of a large storage capacity (potentially utilizing network based storage) and sharing and playback over a computer network can be quite convenient for analysts. This is the latest upgrade to the MAVCapS and will be discussed in more detail in the Digital MAVCapS section of this paper.

3.4.4 Video Distribution Capability Improvements Another important aspect to the MAVCapS was the addition of analogue video distribution capabilities.

In the simplest form, this is erected through direct coaxial cables running from the distribution amplifier in the MAVCapS to televisions in other rooms. This is

suitable for distances up to 20m or so (3). Depending on where the cable needs to go, this can be a quick and simple set up. Expanding on this basic setup, a video switch (or selector) could be used for switching between multiple camera views on the same monitor.

To provide wider distribution capabilities, video baluns were introduced into the MAVCapS systems. These baluns convert the coaxial audio and video cable into unshielded twisted pair such as Cat5 Ethernet cable. This allows two Ethernet ports in separate rooms to be directly connected together (or ‘patched’), and video can be transmitted through the network cabling. This technique can allow colour video to run up to 300 meters (3). With this system, the Ethernet ports must be directly ‘patched’ together as the signal cannot run through any computer networking switches or routers. However, there are also digital solutions now becoming available that can perform the same function but over a TCP/IP network.

Beyond the simple and practical analogue distribution options previously mentioned, a wide array of commercial systems are available, designed for commercial broadcasting usage or large CCTV video surveillance installations. One advanced solution utilised by LOD was the implementation of a commercial video matrix switching system developed by ProVideo (4). Baluns were used to transmit the video from MAVCapS in multiple locations in the experimentation facilities to the video matrix switch. Then the numerous audio video outputs from the matrix switch were transmitted to TV monitors around the facility, again using baluns. Located with the TVs were selector panels, also connected to the matrix switch via direct Cat5 cabling. These allowed users to select any of the video feeds from any of the MAVCapS to view on that TV. It also allowed a number people to view the same feed from a MAVCapS. This proved very useful for the analysts and experiment coordinators in viewing the experiment proceedings. A significant drawback was the large amount of setup, configuration, and troubleshooting time, particularly in patching all the Ethernet ports to the matrix switch location (typically near the patch panel in a server room). As such, this setup tends to be used only for very large scale experiments in Puckapunyal where it was considered an important and worthwhile capability.

Another video distribution capability used occasionally was a hardware Video Over IP device for transmitting live video from Puckapunyal to DSTO Edinburgh over the Defence Experimentation computer network. This consisted of a small hardware device with four coaxial video inputs, an Ethernet output, and an embedded web server. In this configuration, a live stream of any of the four video sources could be accessed by simply browsing to the devices IP address. Quality was limited by the bandwidth of the network connection but it was useful for some applications.

3.5 Digital MAVCapS The latest upgrade to the MAVCapS, known as the Digital MAVCapS, utilises a digital recording component and digital distribution system. The remaining analogue components such as the cameras and microphones have been retained for cost reasons.

The digitization of the capture and distribution components were done through the use of a rack mount computer in place of a VCR or DVD recorder and an LCD computer monitor in place of the TV monitor. The PCs utilise commercial video capture cards to record the video onto internal hard drives. This video can also be streamed live over a computer network if available.

Hardware wise, the consumer video capture cards that were initially tested contained hardware MPEG2 encoders that aimed at offloading much of the video encoding from the computers CPU. However, it was found that the tested cards introduced a one to two second delay in the video appearing on the computer screen. This made it extremely difficult to control and direct the PZT cameras accurately using the joystick controller. For this reason, commercial capture cards that do not use hardware encoding were chosen. Thus, the PCs CPU needs to do the encoding itself, but modern dual core and quad core CPUs are easily capable of this in real time.

Software wise, LOD has utilized Microsoft Windows Media Encoder (5) on the MAVCapS PC, which is a capable and free Windows based software solution. This, when combined with Windows Server 2003 or 2008 running Windows Media Services (WMS) proved to be a relatively simple to configure solution for live video streaming over a computer network. Other potential distribution software, such as Adobe Flash and RealNetworks RealMedia solutions were explored but were costly and more suited to larger commercial audio visual distribution solutions.

Compared to previous LOD solutions, the digital MAVCapS has an advantage in both the audio visual capture component, and the distribution component. The system is capable of recording several weeks worth of video and offers potential advantages in transferring and storing the video. Data can be transferred onto servers and made available for easy access after the experiment.

On the distribution side, digital video distribution over computer networks has many significant advantages over analogue distribution. These include the potential to significantly ease the setup and cabling for the video distribution (particularly compared to the matrix distribution switch system). Computer streaming also allows video to go to locations previously too difficult to lay direct cable to. Any computer on the network could potentially be used to view the live video feeds, removing the need for dedicated TV monitors.

The disadvantages of digital distribution lie in the potential technical difficulties in setting up and configuring the digital video streams and networks.

Often dedicated servers are required for the digital distribution and experienced staff may be required.

Since initially implementing the Digital MAVCapS solution, continual refinements and improvements have been implemented within LOD. Some of the key improvements have been in the area of usability. For example, a new graphical user interface for windows media encoder has been developed to make the windows media encoder software simpler to use and better suited to our needs. On the distribution side, a simple webpage has been developed that displays the floor layout of the experimentation facility. Users on the computer network can simply select the relevant rooms to see a live stream (if a MAVCapS is present, the streaming services are active and the user has appropriate access rights).

Current development within LOD to expand on these capabilities include the ability to remotely control and steer the PZT cameras when watching a live stream, and automatic cataloguing capabilities to allow easy sourcing and playback of video post experiment.

3.6 Permanent Wiring of Facilities After years of regularly setting up and wiring up a room for video recordings and then removing that setup post experiment, permission was granted in 2003 to permanently wire the main experimentation rooms in the facilities in Puckapunyal and in LOD in DSTO Edinburgh and permission was granted to do so. Permanent wiring was run through the ceilings of each room allowing a MAVCapS, cameras and microphones to be quickly installed or removed. This significantly cuts down on the amount of time required for the setup and pack up for each experiment. Custom cabling was no longer required for each experiment.

To ensure flexibility in room layouts, cabling was run for cameras to 12 points spread across the ceiling and 10 points for the microphones. This allows cameras and microphones to be placed in strategic locations around the room, depending on the usage of the room in the experiment. Typically within each room, 5 microphones are used for the audio, 2 PZT dome cameras for controllable video, and a fixed camera with a wide angle lens in a corner to show the whole of the room on video.

3.7 Audio Visual Capture from Other Sources One capability that has been regularly been required when doing audio video recordings to support experimentation is capture from devices other than cameras and microphones. These have typically been images on computer screens or computer audio.

To capture images from computer screens (such as presentation or wargames), devices called scan converters can be used to convert VGA type computer signals into PAL signals for output over coaxial cable. These scan converters typically sit between a computer and monitor and split off a PAL signal while the

computer signal passes through to the monitor. One limitation of these is that the resolution of the PAL video is limited to 768x576 which is much lower than computer resolutions. Generally, PowerPoint presentations can be captured successfully with better quality than pointing a camera to a computer screen or overhead projector screen, but it may not be suitable for recording detailed computer based wargames or simulations.

Occasionally, there is a need to record sounds from a computer during the experiment as well as the video captured from the room. For example, simulation radio network software such as ‘TeamSpeak’ (6) where it has been desirable to record these radio communications with the video from the cameras in the room. One solution that LOD has used successfully is to connect the headphone jack from the computer into the left or right audio channel and microphone outputs are recorded on the other channel. During playback, the analyst can select which type of audio they want to hear by adjusting the audio balance to the left or the right.

There have been occasional requirements to record what a participant is doing or looking at. To capture this, LOD have mounted pinhole cameras on baseball style caps or military helmets. Pinhole cameras are a variation of the typical fixed CCTV cameras but they are very small cameras with a small aperture (hence the term pinhole). These head mounted cameras are an affordable option to consider for experiments and simulations, as well as field trials and operational test and evaluations.

4. CONCLUDING COMMENTS The topics covered above highlight some of the benefits of audio visual capture and distribution in support of experimentation, wargames, and analysis. It also presents some of the tools and techniques that have been used by DSTO. The technologies utilized are relatively simple and affordable, and they are possible to implement without the need for a permanent commercial CCTV system.

When implementing a solution, a strong focus should be put on the true requirements for a given situation in terms of recording and distribution. Early involvement in the experimental planning process is recommended to develop these requirements. Then the options available in terms of current infrastructure, available equipment, funding, skill and staffing can be assessed. The examples given in this paper may help in understanding what to consider when formulating requirements, and options presented and lessons learned may assist in determining how to meet those requirements.

Beyond developing an audio visual system, considerations need to be made for how to manage the data once recorded, and these topics are addressed in a further SimTecT paper entitled “Data Capture And Information Management To Aid Experimentation And Analysis” (7).

REFERENCES 1. Alberts, D. & Hayes, R (2002). Code of Best Practice

for Experimentation. DoD Command and Control Research Program - CCRP Publications Series: Washington, pp. 99

2. Grant, D (1995). Doug Grant’s Guide to CCTV for the Customer. GST: West End, pp. 43

3. Jaycar Electronics Pty Ltd. Installing Your Own Closed-Circuit TV System. Jaycar Electronics: Silverwater

4. ProVideo Australia, www.provideo.com.au, 2009

5. Microsoft, www.microsoft.com/windows/ windowsmedia/forpros/encoder/default.mspx, 2009

6. TeamSpeak, www.teamspeak.com, 2009

7. Crase, S & Bennett, B (2009), “Data Capture and Information Management to Aid Experimentation and Analysis” SimTecT Conference 2009, Adelaide Australia, 15-19 June 2009.

ACKNOWLEDGEMENTS The author wishes to acknowledge the support, advice and guidance of Bruce Bennett and the late Adrian Catford from DSTO. Recognition also goes to Brett McLindin, Ronald Harrod and WO2 Peter Horley for the development of some of these capabilities. The author also wishes to thank the program sponsors including AEF and LWDC for their continued support.