RTVE's technical operation for TOKYO 2020

New Olympic and Paralympic games are about to begin: Tokyo 2020. And RTVE has now the technical infrastructure ready to cover this world sporting event. With a programming that will include 400 hours of first broadcast through La 1, Teledeporte, RNE and their new digital platform, RTVE Play.

RTVE's technical deployment will be carried out thanks to the work of a small group of engineers and technicians who travelled to Tokyo and are already carrying out the tests on the equipment installed there, as well as to the work of many other technical professionals who from the Barcelona (San Cugat) and Madrid (Torrespaña) headquarters will configure, operate and provide 24/7 support to the communications systems and all the broadcast and information technology equipment involved in the audiovisual production chain designed specifically for the Games. Part of the electronic equipment to be used was pre-installed and configured in Madrid

By the RTVE technical team, in July 2021

during the month of May. And in mid-June the equipment was sent to Japan by air.

The remote set that has been set up in Tokyo, located opposite to the Olympic Stadium, will be the main hub from which an important portion of La 1's programming will be presented. Remote production will be carried out from the A4 studio in Torrespaña with the cameras from the Tokyo set. Other Olympic programs broadcast in Teledeporte will be produced from the E3 studio of the Sant Cugat Production Center. Radio Nacional de España from its studios at the Casa de la Radio in Prado del Rey will focus on the Games with in order to narrate live all the medals won by the Spanish athletes. And as usual, the rtve.es website will feature a multi-device offer for the Olympic Games. On this occasion, the games will coincide with the launch of the new RTVE Play platform, an evolution building on RTVE’s Alacarta platform.

The transport of all the signals and the multiple technical services between Tokyo and Madrid or Barcelona will be carried out through Nimbra nodes. These are linked by OBS on two separate 1G Ethernet (820 Mbps payload) bidirectional circuits up to their London and Frankfurt POPs. From those POPs to the RTVE center in Madrid, the 1G trunks are served by Eurovision Services (EBU) through their FINE network. They are not redundant circuits since the services they carry are distributed. In the event of failure of one of the circuits, a redistribution of priority services is programmed by the circuit that remains operational.

The 15 unilateral MDS (Multi-channel Distribution Service) channels that OBS distributes encrypted as allocated by the Eutelsat 7 and Eutelsat 10 satellites will also be downlinked through the Torrespaña parabolic antennas. Internally, once demodulated, by RTVE's own contribution network, all the signals necessary for production will be extended to Barcelona.

As described below, RTVE's public service undertakes a complex technical operation through which the images, sounds and values of the Olympic spirit will be conveyed to the spectators. On top of this technical deployment, hopefully the presenters and editors of the programs will narrate many successes of our Olympians and Paralympians and the exciting award ceremonies of their medals.

RTVE's MCR has been installed in a dedicated space at the IBC (International Broadcasting Center). From there, all the signals and technical exchange services involved in the operation of the Olympic Games are managed. The MCR will receive the OBS IP VandA 1.PK14 service with live signals from the

ceremonies and competitions for all Olympic disciplines. Two 10 Gbps streams -main and redundant- will be received.

IP VandA is a streaming of up to 48 multicast signals that meets the SMPTE 2022-2 and SMPTE 2022-7 video distribution standards. The HD 1080i/59.94 video signals are encoded in H.264 High@Level 4.1a 22 Mbps format with a GOP size of 32 frames. And international sound L/R and AAC LC (5.1) audios are encoded in MPEG1 Layer 2 2/0 at 192K and 512K bit rates, respectively. All signals are multiplexed and encapsulated over RTP (Real Time Protocol). The IPVandA distribution switches are managed by OBS through an SDN controller and the stream is then pushed to the Appear X platform installed by RTVE in its own space at the IBC.

Appear's X platform handles the IP convergence of Digital TV. In a modular chassis with built-in high-availability features, the platform has all the necessary interfaces: IP, video in classic baseband or over IP, modulators, demodulators and all processing capabilities such as encoding, decoding, encapsulation and decapsulation SRT, ST2110, ST2022 in a fully integrated way and with a high processing capacity. In this case, Appear's X20 allows RTVE, on the one hand, to receive the 48 DX from the IPVandA service on both the main and reserve 10Gbps fiber and reconstruct the transport frame in a totally seamless way under the SMPTE 2022/7 standard and on the other, it allows the decoding of the 4 services corresponding to the competitions of interest at any given time in an equally redundant manner and totally dynamically- chosen by the operator, thanks to the intuitive graphical user interface offered by the X platform. The decoded signals are delivered to the MAM card in the Nimbra node that encodes them under JK2 and delivers them to the OBS circuits towards Spain.

Although the Appear X platform has excellent features in terms of redundancy, the top

RTVE's MCR at the Tokyo IBC

priority is to ensure maximum availability of the images of the competitions. For this purpose, two Sapec Laguna octal processors have been connected in 10G, which in case of failure can take over the decoding of the IP VandA signals.

The signals decoded by both Appear and Laguna are delivered in HD-SDI baseband to a Utah Scientific HD-SDI matrix. This longstanding, proven matrix, which has already served in previous editions of the Games in Rio and London, routes all the bidirectional exchange signals between the MCR and Madrid and between the MCR and the remote set located opposite the stadium to the Nimbra network. The matrix also interconnects a multiscreen for local monitoring of the signals and a measurement TOM for quality control.

A 1G Ethernet link through which OBS connects the MCR with the so-called Announce Platform position contracted by RTVE inside

the Olympic stadium. Through this trunk, an ENG signal from the stadium will be received via IP and the program return PGM signal will be sent to it with the N-1 audio + commands. For this, both the Announce Platform and the MCR have two Sapec Laguna processors that encode and decode the HD-SDI signals in IP bidirectionally with the audio embedded between both points. The MCR's Laguna is connected to the matrix for the routing of signals with Spain.

An RTVE Nimbra node provides all the necessary communication services between Torrespaña, Sant Cugat and the MCR and between the latter and the remote set opposite the stadium. The node features MAM cards for the Tx/Rx exchange of video signals encoded under JK2 at 150 Mbps, cards dedicated to AESEBU audio Tx/Rx and cards for internet services: linking of the Artist Riedel intercom nodes, IP telephony, corporate services, digital newsroom, CIS (Commentator Information System), ODF (Olympic Data Feed) Management, etc. This node, through redundant 1G links, is connected to the remote set's Nimbra node and to the OBS Nimbra node for connections with the London and Frankfurt POPs.

For the presentation of broadcasts from Tokyo, RTVE has set up a set there that has a large window with a direct view of the Olympic stadium and LED interior lighting. To control outdoor light exposure and deliver the best image quality, Rosco View polarizing filter systems have been fitted to the window and set cameras. By rotating the filter in the camera, the degree of cross polarization between the two is changed and the backlight is balanced with the lighting of the set. The use of these filters only entails a reduction of one diaphragm stop.

The programs in Tokyo will be carried out as remote production from studio A4 in Torrespaña. Four cameras have been arranged at the Tokyo set (one of them as a back up), Grass Valley LDX82 Premiere model with their XCU Worldcam base stations that will be permanently installed in RTVE's Castilla - La Mancha Territory center after the Games. These cameras can work in HD 1080i/50 format since the organizers of the Games provide the 230V/50Hz power supply for the set. A Calrec Brio audio mixer and an IP system for the Autoscript text presentation (teleprompter) have also been fitted. These resources belong to the Prado del Rey studios. In the set, the CUE is connected via Ethernet to a switch that in turn is connected to Nimbra with a bandwidth of 20 Mbps. The presenters' tablets are also connected to Nimbra, since, although they run

via WiFi, the access point is directly connected to the Nimbra. In this way the need for a VPN is avoided. WiFi internet connectivity would be a backup, like the 4G router.

The installation is completed by a Riedel Artist intercom node for N1 audio communications, commands and GPI for Tally cameras and cable or Phonak WisyCom in-ear devices for presenters. On the set, large-format LED monitors will show image content sent from Torrespaña.

Through the Calrec console, a sound operator manages the signals received from the Sennheiser 6000 digital wireless microphone, the N-1s and orders that arrive from Spain through Artist for the listening of presenters, the set's PA system, the sending of news and sports audio to the embedders who insert them into the cameras' HD-SDI signal and the back up sendings to Spain from the set microphones

RTVE's remote set in Tokyo located opposite the Olympic stadium and electronic equipment installed on the set

through the AES-EBU cards in the local Nimbra node.

The Nimbra node in the set has MAM, AES-EBU and Ethernet cards; it supplies all the necessary services for the production and the network trunk of the Artist intercom node. Through this Nimbra, the set's screen signals, the CIS, are received, a video signal is exchanged with the IBC, the signals from the cameras are sent directly to Madrid as well as the return of the PGM program from the Torrespaña studio and the broadcast returns of La 1 and Teledeporte. Camera signals are encoded under JK2 at 150 Mbps and sent to the Madrid studio with a very low latency. The program return and the videos for the set's screens reach the set also encoded in JK2 at 150 Mbps. Broadcast and Announce Platform returns not requiring as much quality arrive encoded at 100 Mbps.

Additionally, in order to guarantee the continuity of the program from Tokyo in the face of any issues in the OBS or EBU circuits, a Net Insight encoder model Nimbra VA 225 has been installed as back-up to send the 3 camera signals with the audios from presenters embedded, encoded in H.264 at 8Mbps and channeled to Spain through a dedicated 20Mb capacity FTTH circuit. This encoder is integrated into the Nimbra Vision system deployed on RTVE's contribution network, through which the device and the services assigned to it can be managed and remotely monitored.

For the presenters at the set, a small 3-computer newsroom has been set up and connected to the iNews systems in Madrid and Barcelona. A network printer is also available. And the teleprompter is connected to iNews so as to be able to write in the step outlines as if they were in the TVE's newsroom. They will even be able to insert labels in the news of the step outlines. The set contains 3 Surface GO2 tablets connected in the same way as the computers so that the presenters can follow the list of the program or the news program. Set connectivity is carried out via the internet and access to AVID's production network via VPN tunnels. There is also a 4G router as a backup connection system in case of internet service failure.

As indicated in the introduction to this article, Torrespaña is responsible for downloading the OBS MDS transmissions from the satellites. These are demodulated by NS2000 Novelsat equipment featuring temporary DRM licenses for signal

decryption. Each demodulator delivers an MPEG transport stream with three services in ASI format. By means of 5 demodulators, as many ASI signals are obtained and shared internally by Nimbra with Sant Cugat. Both Madrid and Barcelona have H.264 MPEG-4 4.2.2 decoders to obtain the Olympic signals in baseband video.

On the other hand, the circuits of the FINE network reach the EBU's Nimbra node in Torrespaña. The two 1G circuits of this node are connected to the Nimbra node of RTVE in International Control. From there, all the services that are transported are managed by RTVE's technical staff in charge. The IP VandA signals selected at the Tokyo IBC through the Appear platform arrive in their original HD 1080i/59.94 format. To convert the frame rate to the European 1080i/50 format, the signals have to go through high-quality converters with the Alchemist Grass Valley motion compensation algorithm and are then sent to a matrix for distribution to the various internal destinations. The signals from the cameras at the set are also received, which when working natively in 1080i/50 do not require any conversion and are sent directly to the mixer at the A4 studio.

Signals coming from Tokyo with international sound (either through the MDS or through Nimbra) are forced to pass through various call centers where the various specialists will comment on them, thus generating a video signal with comments in Spanish. Projections are for three call centers in Torrespaña and another four in Sant Cugat. The respective central controls in Madrid and Barcelona will be in charge of composing the signals with international sound with the locutions, generating signals with embedded audio ready to be delivered to the studios or for feed, depending on broadcasting needs.

At the Torrespaña and Sant Cugat controls, signals will also be received from up to 6 Ikusnet IP transmitters from Prodys taken to

Japan for news coverage at the various Olympic venues. Each ENG transmitter has 6 3G/4G modems. The signal from an ENG camera is encoded in H.264 and sent live via IP or stored as a file and then sent to Torrespaña via FTP.

All the aforementioned signals will be available both for feed and for use at the radio studios in Prado del Rey, for production of rtve.es, for work at the digital newsrooms and for production of the relevant programs at the respective studios: A4 in Torrespaña and E3 in Sant Cugat. Specifically, in the A4 studio that will carry out the remote implementation, the provision of audio deembedders has been expanded. so that the set of signals that it must handle is distributed as follows:  3 cameras from the

Tokyo Set, two of them with embedded audio

 4 signals for booth feed in Torrespaña, to broadcast replays and recordings  3 signals from the Digital

Newsroom, for broadcasting news edited by sports journalists  6 signs of commented sports  2 signals from the E3

Studio in Barcelona, to be sent* to the monitors at the Tokyo set  2 signals from the A2

Studio (Daily News) in

Torrespaña, for sending** to the monitors at the Tokyo set  2 signals for 4G backpacks taken to Tokyo

* when the E3 studio in Sant Cugat carries out most of the broadcast

** during the broadcast of the news

Torrespaña's A4 studio has temporarily expanded its technical equipment to facilitate these tasks. Ancillary to the aforementioned deembedders there is an extension from 12 to 16 lines from the central control matrix, 8 additional inputs in the video mixer and multiscreen, a new character generator with features that will be later described, and the generation of a series of different PGM output signals from the studio, depending on destination: Return PGMs (different for Tokyo, for Studio A2 and for Studio E3) that are added to the usual PGM signal, plus the usual spare PGM circuits.

Additionally, the studio, since production will be carried out remotely, also has computer platforms for intervention in programs featuring remote guests through the Skype and VMix applications that had already been widely used during the toughest stages of the pandemic.

Due to the time difference with Japan, the bulk of the sporting activity will take place during the Spanish early morning and midday. Teledeporte and La1 will allocate the events between them in order to be able to offer the most number of live broadcasts. Teledeporte will have a 24-hour broadcast with Olympic content, whether live or recorded. And La1 will simultaneously broadcast live competitions from 0 am to 3 pm, the time at which

Daily News 1 will begin, which will also include a significant amount of Olympic and Paralympic information.

To manage the broadcast on both channels, cooperation among the A2 and A4 studios in Torrespaña and the E3 studio in Sant Cugat will be required, as well as the from the Tokyo set. In them, all the production of the Olympics will be made in 5.1 audio and will be broadcast in this format on La 1 HD and the TDP channel.

Torrespaña's A4 studio will centralize most of these operations, switching between different work modes according to the time slot:  From 0 am to 9 am it will run the Teledeporte channel with live broadcasts.

 From 9:00 a.m. to 3:00 p.m. it will run the broadcast of La1, in which it will include the Tokyo set, with all its resources and live presenters.  From 3pm to 4pm the broadcast of La1 goes to studio A2 in Torrespaña, but studio A4 remains operational to carry out the production of the

Figure 1.

Tokyo set during the broadcast of the daily news. At this time, the Tokyo set screens will receive signal from studio A2, routed from studio A4.

From 4pm to 5pm, a summary of the day will be broadcast on La1 from the E3 studio in Sant Cugat. For this broadcast, the A4 studio will continue to operate the remote production of the Tokyo set as during the news broadcast, in this case routing the signals from the E3 set's screens to the Tokyo set.

The E3 studio in Sant Cugat, recently renovated and equipped with modern production technology over IP based on the ST2110 and ST 2022-7 standards, will feed the continuities of La 1 and TDP in different time slots. For the occasion, a new set is inaugurated, which includes largeformat LED videowall screens that will display what is happening. It will have a supply of 5 cameras with tripod, 1 crane, 2 mini cameras and 1 robotic camera. The following signals will reach this studio:  7 signs of commented sports

 1 sign from Torrespaña's

A4 studio, with the signal produced from the

Tokyo set

 6 signals for 4G backpacks deployed in Tokyo

 An EVS IPDirector system for broadcasting replays and recordings

 8 signals from the Digital

Newsroom, for broadcasting news edited by sports journalists

To label the programs, new templates have been designed specifically for the Olympic Games. In the case of the broadcasts made from the A4 studio in Torrespaña, these templates are integrated with the iNews System and the Chyron CAMIO server. In the production stage, labels are inserted in the iNews news by using the LUCI plugin and the CAMIO system (integrated with iNews). The newsroom only has to select the Olympics context in CAMIO through the LUCI plugin and drag the label to the relevant piece of news. The entire system is conveniently configured so that the loading of labels is carried out on a Lyric X device. In broadcast, an ISQ label automation system has been deployed that will receive the list of labels from iNews and will control the Lyric X in order to air them. In the case of Barcelona, the labeling solution is a similar one, but is based on the Avid Graphics (Orad) graphics system. That is, the integration is with Maestro/Command. The labels are entered through iNews but using the Orad plugin. At broadcast stage, they are launched from the Maestro client.

Intercommunication of the entire operation is fast since all RTVE facilities have the Riedel Artist intercom matrix system. The intercom installation is based on the trunking system, which enables communication between different venues through the nodes of the Nimbra contribution network. The link among venues is

made by several IP audios. In this way, the three venues -Sant Cugat, Torrespaña and the Tokyo Set- will be united. This makes it possible to properly manage coordination communications, orders and audio returns for all the intercom panels, cameras, in-ear devices installed in the multiple workstations involved in the operation of the Games: studios, central controls, MCR and Tokyo set, presenter booths, etc. All nodes are interconnected VoIP through RTVE's Nimbra contribution network. In addition, as already indicated, this facilitates Tally and GPI/O services. Through Connect IP devices it allows VoIP transport from the commentator positions at the different Olympic venues.

In the case of the IBC, as it only needs two intercom panels for coordination, the connection is made through a Riedel Connect Duo device that allows IP transport of a communications position.

Figure 1 summarizes the entire network set up to service the operation that RTVE's technical areas have designed for the Tokyo Olympic Games.

Due to the restrictions put in place for these Games, the number of people who can travel to Japan is very limited, so the better part of the programs will be carried out from Casa de la Radio. To achieve this, editors will have the live video signals of the various sporting events in their booths. These signals reach them internally from Torrespaña through the Nimbra contribution network. Through it, 8 HDSDI video channels have been set up with the contents concerning competitions of interest managed by Torrespaña. The signals will be distributed by means of a KROMA video matrix and by Ethernet cabling (in HDBaseT), to the main studios and recording booths of the Casa de la Radio. During the broadcast of Olympic Games special programs, commentators will have, on the television sets in the studios, the video signals of all the events that are being held at that time, so both broadcast and comments will be will always be performed in real time. 6 editors and 2 sound technicians are in Japan for radio services. Technicians will be mainly in charge of the production of the programs, audio editing and technical support to the editors. However, due to the special circumstances surrounding these Games, it has been considered essential to give the editors greater freedom of action. For this reason, backpacks have been prepared so that they can easily transport the technical means necessary for conducting interviews and live interventions from the different venues. Each of them will have a laptop,

prepared for audio editing and recording with the Dalet production system, a Quantum Lite audio encoder, for live interventions via 4G or WIFI, as well as a mobile phone, microphones and a pole for interviews. The aim is to ensure that, with a minimum number of people allocated to the venues and taking into account mobility limitations that they are going to face, information coverage as close as possible, to the broadcasts of previous Olympic Games can be achieved.

The production of up to 400 hours of live broadcasts for these Games makes it possible to form a package that includes the broadcasts of La1, Teledeporte and up to four additional live channels, which will enable to offer up to six simultaneous live shows of sports events through rtve.es. Broadcast of thematic channels during the Games is also being considered. The RTVE Play video-on-demand platform has just released its trial version and will improve through user experience before its official launch in autumn. Digital coverage of the Olympic Games will reach almost all digital video consumer platforms, with special emphasis on YouTube, where summaries will be offered with the highlights of each day's competition.

Technical operation of the website can be split into four different processes:  Contribution From the

Torrespaña Control, signals are distributed in

HD-SDI

Access page for the recently launched RTVE Play VOD platform

 Multi-quality transcoding and storage in segments from 5s to 10MB/s in Full-HD quality

 Packaging and broadcast of the live signal through CDNs (Content

Delivery Networks).

 Cutting and publishing the VOD from the recordings.

For live production and for the catch up, the HDSDI signals distributed from the Torrespaña Control are used, which are transcoded in Transport Stream by the HLS protocol and 5second segments that reduce latency and allow the information to be stored with maximum quality. Processes are used for broadcast from these live recordings in real time, as well as for subclipping (creating cuts directly from the live show) for the newsroom. The packaging, which is mainly done in HLS format contains four different quality standards: 1080p, 720p, 576p and 360p, all of them in perfectly aligned MP4 format. A DASH format is also generated, mainly to meet the needs of connected TVs and HbbTV.

Two pairs of stereo audio are used to manage the audio signals. Pair 1 with ambient sound of the event + commentator. And Pair 2 only with ambient sound. As distributed HDSDI signals carry two pairs of audios, the first one with the mixing of ambient sound plus the commentators' speech, and the second only with ambient sound, this allows a single recording with both types of sounds, so that at a later stage an editor -through his own

editing tools- can generate pieces with the required audio. For online production, the complete ones (complete tests), publication with the comments by the announcers is needed, but for short videos only the ambient sound is used in all instances, since otherwise the end user could become confused. For longer summaries, specific comment may be added. Finally, these pieces are uploaded to the rtve.es CMS to be made available to users.

From a more technical perspective, production of live show has the following work scheme:  The HD-SDI signals distributed from the

Torrespaña Control are injected into two systems, a Makitos (Haivision) encoder farm and the Golumito platform from the Golumi company.

 The Golumito system generates the 5-second segments with the required quality levels. It also generates the HLS and DASH playlists in four qualities, and publishes all the information to the CDN Publishing Netstorage (Akamai).

 Live images are generated every 10 seconds for distribution of live thumbnails.

 The Golumito system generates three versions of encapsulation in real time:

-Five-segment encapsulation for low capacity devices. -DVR encapsulation (including segments from the last three hours) -StartOver encapsulation. For this, each live show taking place on each signal is uniquely identified, and its specific playlists are then generated.

Regarding the VOD (catch up), the process is performed as follows:  The editor connects to

Golumi's Golumito system, selects the portion of the recording to use, and uploads it to a video environment. The pieces intended for export, as well as the audio band (voice-over or ambient) are chosen and finally exported to the CMS.

 The Golumito system composes from the selected segments a package with MPEG4 media, the metadata and the audios. As Golumito actively listens to events coming from TVE's continuities (Harris), it includes the metadata that it sends plus any additional data chosen by the editors, as well as other stored audios such as audio description, original version, etc.

 The packaging is fed into the CMS through RTVE's

Ingester system.

 For the distribution in CDNs, an on-the-fly segmenter and the Unified Streaming mp4split, are used and the segments are persisted in the dis-

tribution CDN (Fastly or Level3).

In the specific case of HbbTV, for compatibility with TV receivers prior to the 1.5 standard, the Golumito system generates for each direct a streaming in RTMP format that is fed into an external provider (Cinfo) that generates and distributes a Transport Stream from the source signal to end users.

In addition to the abovedescribed main management system, RTVE has a second backup ingest subsystem, the Cires21 LMS. Using the same HD-SDI signals through the Makito encoder farm where IP signals in multicast format are captured by the LMS to broadcast to the backup endpoints.

For this Olympic event, RNE are going to make a special effort to report from smart speakers, especially on Alexa. Two years ago, RNE launched the RNE skill in Alexa, in which you can consume the content that RNE produces, this being the product from which you can follow the Games. A workflow has been created to guarantee the immediate publication of the contents so produced, integrating the radio’s Dalet production system with the CMS of rtve.es. In Dalet an export procedure has been created by which the content selected by the editor passes to an Interactives FTP system. This integrates an active listening system that, when faced with new content, launches a technical ingest and standardization process.

The rtve.es ingest system carries out an audio standardization by adjusting the maximum and minimum thresholds (difference between audio maximum and minimum levels measured in decibels), to later encode the standardized audio in mp3 at 128Kbps bit rate. Then the media and its metadata, the title of the audio and the category of the program, are uploaded to the interactive CMS. Finally, from the interactive CMS, the audio is published through a REST API and the mp3 is published through a CDN (Fastly).

For its part, RNE's Alexa skill organizes the contents into three sections: agenda, Spain's summary and summary of the day. Technically, the skill has been implemented by using Amazon lambdas that acquire the metadata using the REST API of rtve.es, and if the user so requires, content is played by accessing the media published on the CDN. 