Related article:

Part 1 of the BEITC story covered TV engineering sessions on Saturday and Sunday, 5 and 6 April. Part 2 covers relevant Monday and Tuesday sessions. There were many more BEITC sessions, but the following are some of the most relevant for Broadcast Bridge readers.

Monday 7 April 25

The first Monday session of interest was “Proactive Observability for Video Over IP,” presented by Adi Rozenberg, CTO at AlvaLinks. 

Proactive observability is all about injecting packets into the network for discovering the available paths from source to destination and continuously testing the network impact on the packets traveling across these paths. The injecting is done continuously to be on top of the route changes and response slow-down and to identify which hop may be causing issues. In parallel, packet-by-packet relationship information is collected: delay variation, latency from the source, reordering, loss of packets and more.

The sending probe starts by trying to reach a receiver probe IP address. The probe sends packets with an increasing Time To Live (TTL) value. The probe monitors every returned packet with Internet Control Message Protocol (ICMP) time exceeded. Then, determine the round trip time (RTT) by comparing the internal timestamp with the time the packet was received.

Unlike standard observability that just collects data from elements and passive probes, proactive observability involves injecting small amounts of traffic into the network to discover the network routes and to support the continuous measurement of the performance of the traffic to gain valuable insights into the experience of the traffic.

B2B Content Distribution Over CDN for OTT and Broadcast Services was presented by David Eisenbacher, CEO/Co-Founder, EZDRM Inc.

By using the Common Encryption Scheme (CENC) and the Content Protection Information Exchange (CPIX) protocol, B2B Content Distribution provides a secure, flexible, and cost-efficient solution for live and on-demand content delivery. This approach also supports advanced features like channel-based access control, real-time revocation, and blackout management.

Satellite-based content distribution, such as BISS-CA (Basic Interoperable Scrambling System Conditional Access), has long been the backbone of broadcasting. OTT streaming is more versatile and modern than the linear MPEG-2 Transport Stream (TS) format because it leverages advanced technologies designed for the dynamic and interactive nature of internet-based delivery. To ensure content protection, CDN-based distribution incorporates robust DRM technologies.

"Integrated Newsrooms with Generative AI: Efficiency, Accuracy, and Beyond" was presented by Punyabrota Dasgupta, Principal Solutions Architect at AWS India, and Maheshwaran G, Principal Solutions Architect at AWS India. The paper explores the agentic AI approach that can be adopted to optimally orchestrate the newsroom workflow end to end. Agentic AI is a type of AI that can operate independently with minimal human supervision.

In a typical fragmented newsroom operation, broadcast, digital and print departments often process story information independently. The result can be different narratives or presentations of the same story. The integrated newsroom model reduces costs and duplications. It streamlines by fostering seamless collaboration across departments. Central to this model is the use of generative AI, which enhances content creation, automates workflows, and personalizes news delivery.

A centralized content hub in an integrated newsroom acts as a unified repository where all media assets, such as video clips, text, audio, images, and other resources, are stored and easily accessible to different departments. This centralized approach eliminates the need for separate repositories for each platform, ensuring that all teams, whether working on TV, digital, or print media can access the same content simultaneously.

"Two-Pass Encoding for Live Video Streaming" was presented by Mohammad Ghasempour, a student at Alpen-Adria-Universität, Hadi Amirpour, and Christian Timmerer with ATHENA. 

Traditional live video streaming typically relies on single pass encoding due to its low latency. However, it lacks video content analysis, often resulting in inefficient compression and quality fluctuations during playback. Constant Rate Factor (CRF) encoding, a type of single-pass method, offers more consistent quality but suffers from unpredictable output bitrate, complicating bandwidth management. In contrast, multi-pass encoding improves compression efficiency through multiple passes. However, its added latency makes it unsuitable for live streaming. This paper proposes OTPS, an online two-pass encoding scheme that overcomes these limitations by employing fast feature extraction on a downscaled video representation and a gradient-boosting regression model to predict the optimal CRF for encoding.

The process begins by extracting a set of features from a downscaled version of the input video segment to minimize the computational complexity of the process. Using these features, along with the original video characteristics such as resolution, framerate, bit depth, and color format, OTPS predicts an optimized CRF to guide the encoding process. Machine learning techniques have made video streaming more efficient and have been successfully integrated into video streaming applications.

Consequently, alongside video content features, these parameters are included as model inputs. To ensure precise prediction, a dedicated model is trained for each target bitrate. Because CRF encoding allows for floating-point values, the prediction process is structured as a regression problem to minimize the output error. To this end, a gradient-boosting regression model is used. This ensemble technique is known for its high accuracy and the ability to capture complex data patterns.

"Verifying Video Signals Using Computer Vision and Machine Learning Techniques" was presented by Saurabh Jain, Director, Strategic Partnerships at Interra Systems.

In general, video quality analysis can be divided into three main categories: signal measurements, adherence to legal requirements and guidelines, and perceptual quality. For measurements, every pixel value is examined and analyzed to obtain video metrics and the widths of black bars, types of color bars, high dynamic range (HDR) optical light levels, and composite signals, and to log YCbCr and RGB range values. Higher-level operations such as corner detection, edge detection, contour finding, connected component analysis, motion vector estimation, and various frequency analyses (including Fourier transform, discrete cosine transform, Wavelet transform, as well as denoising, are used to measure quantities and validate them according to legal standards.

Other measured artifacts include blockiness, blocky dropout, jitter or sync loss, defective pixels, color banding, and RGB/CIE color gamut spaces. Interra Systems is working to enhance and broaden the detection power of video verification using data-driven strategies. These innovative methods will concentrate on uncovering the fundamental spatial and temporal patterns within video data.

“Tower Site Safety: A Comprehensive Approach to Risk Management” was presented by Clark Lindstrom, Director Safety Programs at American Tower Corporation. Tower site safety is critical for preventing accidents and protecting workers from hazards associated with tower climbing, RF exposure and structural failures. Routine inspections help identify potential safety hazards before they become critical. Emergency contacts must be listed at each site.

OSHA 29 CFR 1910.37 outlines emergency exit access requirements, and NFPA 730 provides guidelines on site security and surveillance. Perimeter fencing, signage, and CCTV monitoring should be used to ensure restricted access and prevent unauthorized entry.

The FCC’s Maximum Permissible Exposure (MPE) limits, as referenced in OSHA 29 CFR 1910.97, should be strictly followed. Workers must use personal RF monitors and coordinate power down procedures with antenna owners to minimize exposure. ANSI/IEEE C95.1-2019 provides further guidance on safe RF exposure limits.

OSHA 29 CFR 1910.268(g) mandates the use of fall protection for telecommunications workers, requiring continuous fall protection when working at height. Additionally, ANSI Z359.2 outlines managed fall protection programs, ensuring that safety climb systems and anchor points are regularly inspected.

OSHA 29 CFR 1910.151 mandates that first aid supplies be readily available. OSHA 29 CFR 1910.66 mandates regular inspection and maintenance of personal fall arrest systems. ANSI/TIA-222 provides guidance on structural maintenance for telecommunications towers, ensuring compliance with load-bearing and environmental requirements.

“Optimizing Dynamic Ad Insertion (DAI) for ATSC 3.0 in Low Broadband Access Markets” was presented by Liam Power, Senior Systems Engineer at ONE Media Technologies. Linear broadcast ads have been limited to a ‘one size fits all’ method of delivery for the most part. It was therefore important for the ATSC 3.0 standard to meet or exceed this capability for TV broadcasting.

The ATSC 3.0 standard provides a few options for handling ad insertion with distinct pros and cons to each. XML Linking Language (XLink). In this mechanism, the broadcaster generates a DASH stream with an xlink:href attribute, generally assigned to a period. This is normally done at the beginning of an ad avail.

The DASH specification defines an EventStream element. It essentially forms the DASH version of the SCTE markers common to the Serial Digital Interface (SDI) and Transport Stream (TS) worlds. Events signal the time and duration of various data to the receiver. EventStream comes in two forms: inband, and media presentation description (MPD) -based events. Inband events can be found in the emsg box, with an InbandEventStream element in the MPD to indicate which representation contains the emsg box, while MPD-based events are located fully within the EventStream element in a period of an MPD.

In another method, the broadcaster plays an ad via the broadcast application (BA) at a specific time via the Application Media Player (AMP).

The best path forward is to handle ad insertion with a combination of EventStreams and the RMP workflow. Many more details and requirements were presented than space in this review allows.

Tuesday 8 April 25

"Off-piste 5G in the Broadcast Auxiliary Service Band" was presented by Samuel R. Yoffe, Senior Systems Engineer at Neutral Wireless.

5G can be used to provide flexible, high-capacity and low-latency networks suitable for broadcast content acquisition or delivery, but access to suitable spectrum can be challenging. One of the enablers for private network deployments is shared spectrum licensing, such as the upper n77 band (3.8–4.2 GHz) available in the UK and elsewhere in Europe. In the USA, the Citizens Broadband Radio Service (CBRS) band provides shared access to 150 MHz of radio spectrum in the n78 band in areas where the incumbent user (the U.S. Department of Defense) is not operating. In the USA, the Broadcast Auxiliary Service (BAS) bands mainly operate in ultra-high frequency (UHF) and microwave spectrum, including allocations in the 2 GHz, 7 GHz, 13 GHz and 17 GHz bands. 

One advantage to running a fully SDR solution and removing the reliance on COTS hardware manufacturers and 3GPP specification is upgradability.

“Deploying 5G Broadcast in UHF Spectrum,” was presented by Javier Fernandez, Senior Engineer at Qualcomm Technologies Inc.

Since 5G Broadcast targets broadcasters instead of cellular operators, the most important target spectrum for this technology is the portion of UHF spectrum allocated to broadcast systems (a subset of ~470 - ~694/698MHz, depending on the region). The channelization used within this spectrum can vary between 6, 7 and 8 MHz in bandwidth, depending on the geography (such as those defined by different ITU regions).

After completion of the physical (and higher) layer designs in Release 17, enabling 6, 7 and 8 MHz PMCHs to be supported, in Release 18, broadcast UHF bands (with band numbers characterized by the carrier frequency and associated channel bandwidths) were added to the 3GPP RAN4 specifications.

"Video Super Resolution for Broadcast Television" was presented by Christopher Bird, Cloud Software Development Engineer at Intel, and Surbhi Madan at Intel.

Super-resolution is a technique to scale lower-resolution images or videos to higher resolutions improving its visual quality. Utilizing sophisticated algorithms, including those driven by artificial intelligence (AI), super-resolution can smartly upscale and improve visual quality. Advancements in developing efficient Video Super-Resolution (VSR) technologies can democratize access to high-quality video enhancement for broadcast television, ensuring a superior viewing experience across all platforms.

The paper details the TSENet model architecture built upon the foundation of the Equivalent Transformation and Dual Stream (ETDS) model, which proposes a dual-stream approach for learning higher-frequency (HF) and lower-frequency (LF) components of the input image during the training phase. Each branch is designed to learn different components of the video frames, with the LF branch focusing on low-frequency details and the HF branch on high-frequency details.