Optime

The client incorporated an ultrasound machine into the existing telemedicine kiosk product line and needed to add a low latency streaming solution that would be able to pull the real-time images for the doctors to inspect remotely. Since the patient is supposed to operate the ultrasound probe, the transmitted video stream would need to have the lowest possible latency to accommodate long-distance delays and allow doctors to interact with the displayed images in real time.

We allocated a dedicated team to research the existing product, ultrasound hardware, and its specifications. Our engineers integrated the given ultrasound machine with the WebRTC library to facilitate low-latency signal transmission over long distances.

The client had a rapidly growing number of customers and needed to scale existing systems with a more cost-effective solution. Previously, they had utilized the Polycom service, but it did not support WebRTC SFU architecture. The calls with a growing number of participants were usually moved from Mesh to MCU topology.

However, since employing the MCU architecture is extremely compute-intensive for the server-side the client’s costs were exacerbated, hence the need for a more efficient solution.

We designed a custom media server utilizing Mesh/SFU topologies, which ensured that the growing number of participants was not an issue during transitioning anymore. Employing SFU architecture also meant that server loads would be reduced, hence decreasing the client’s operating costs as intended.

Since the start of the COVID-19 pandemic, patient visits to clinics have significantly decreased and online consultation services have become more prevalent. This forced the client’s telemedicine cart to become less effective as the software on these devices was not integrated into the internal call service of the clinics. Many health care providers were using different WebRTC-based solutions, such as Twillio, Vonage, Vidyo, Zoom, MS Teams, and Google Meet instead of SIP-based services, as their online consultation delivery platforms.

We designed a new system utilizing WebRTC SFU/MCU topologies and built a gateway server to incorporate different video conferencing platforms by reverse engineering latter services. Our hybrid media system was mostly implemented using technologies: C/C++, Linux, Typescript, Node.js, NestJS, React, Mediasoup, Kurento, REST API, and Socket. IO. 

After the upgrade to a 4G network, the usage of traffic by end-users has increased dramatically. Unfortunately, the client's existing GTP analyzer (for data billing) couldn’t handle a new level of traffic.

We utilized Intel DPDK to decrease the latency caused by the operating system’s network stack and implemented a GTP-C parser to correlate a user's mobile phone number with the assigned IP address. The correlated IP address/phone number data was used in the GTP-U analyzer to record data usage per user’s mobile phone, and the analyzed data was stored in a database.

The client requested a full network visibility tool that could analyze IP traffic and identify the issues and their location. The IP traffic analyzer system was supposed to give the client a full picture of their network and the ability to respond to customer requests promptly through call centers.

We implemented a network traffic aggregation system using C and Intel DPDK toolkit, added a traffic classification engine, and created the IP traffic analyzer stack for different protocols such as SMTP, POP3, IMAP, HTTP, various SQL protocols, etc.

The client decided to extend the IP traffic analyzer solution to create header stripping capability for different L2 and L3 protocols, such as QnQ, MPLS, VNTag, GRE, and GTP. Since the original IP traffic analyzer was designed as a pure IP traffic solution, it was not able to fulfill additional needs. However, the client wanted the additional requirements to be developed as a separate solution.

We created an extensible system that could incorporate additional modules for each protocol for the client to be able to further develop custom modules.

The client’s existing multimedia distribution network, with multi-screen video delivery and ABR streaming, encountered enormous challenges to deploy an end-to-end quality monitoring system. They needed a solution that would be flexible enough to deploy at each demarcation point of the operator's network and provide end-to-end visibility and analysis.

We provided an automated quality assurance system for the high-volume video streaming service over IP networks. The system implements a continuous monitoring solution, which conducts per PID analysis at different disjunction points of the operator's network to guarantee rapid fault detection and streamline the rectification process

The client wanted to deploy IPTV service with VoD features in the existing network infrastructure and provide scalable service for the multi-region customer base. The low-latency content delivery needed to be implemented while providing low-cost hardware requirements for the set-top boxes.

We created a low-latency solution that used content caching in main memory, avoiding high latencies of reading from storage. Therefore, the client was able to drive the total storage cost down by employing mid-range HDDs rather than high-end ones or SSDs. We implemented a time shift feature allowing customers to record and replay favorite content.

The client wanted to deploy IPTV service with VoD features in the existing network infrastructure and provide scalable service for the multi-region customer base.

We modified the streaming media server that was responsible for live streaming. The content was transcoded into multiple protocols and multimedia container formats. The solution is optimized with adaptive bitrate streaming, which provides the highest possible quality to users and reduces buffering.

After successfully implementing the first streamer project, the client requested two following projects a Transcoder and a Video on Demand System. The project’s main challenge was to deliver high-quality IPTV service with VoD features and keep deployment costs to a minimum, to easily reach economies of scale and offer competitive price. The second challenge was to develop a scalable solution, which would be sufficient for an increased customer base and handle nationwide traffic loads.

he solution provides low delay/low latency streaming and can handle transcoding up to 11 FullHD channels simultaneously per socket. The solution supports high performance using hardware acceleration based on Intel Quick Sync (using libVA) dedicated video encoding/decoding core, resulting in superior speed and performance.

The project involved a broadcast-grade high-density video processing platform for uncompressed-over-IP function. The device supports video/audio processing, compression, and multiplexing in a single scalable platform. Most common encoding standards like MPEG-2 and H.264 are supported in SD and HD formats.

At Optime we improved platform software meant the device could support more streamlined video processing pipelines as well as offer a smooth and intuitive GUI for management applications.

The project involved a high-density, scalable media processing platform for SDI to IP conversion. 

Our team worked with the client to improve overall software architecture and implement various new functionalities over multiple years of collaboration.

A high-profile sports event needed an ultra-low latency multiviewer for live broadcast production which needed to be built on top of the client's existing media processing platform. 

A new personality was added to the existing platform to enable HD/UHD multiviewer output handling.

We Utilized C/C++ for the integration part with the hardware accelerators in the device.

The client needed a high performance, low latency, small memory footprint, and decentralized data fabric storage system designed to sustain uninterrupted network partitioning.

Our Solution

To create the exact Data Fabric System, the customer needed, our team to design a Storage System. It was specifically designed to enable high throughput workloads; the system had the benefits of:

• High-performance object storage

• Built with throughput requirements of Analytics and Big Data storage

• Cost-effective storage platform

• Secure for long-term storage with guaranteed data integrity

To increase performance and stability we patched several open-source projects such as:

• libuv, Node, js, corosync.

• The core was written in C and the backend in Node

• Js created Node.js binding in C++ to use functionally so users could manage storage through Rest API.

The client needed a fully customized voice-fraud detection solution, which would be able to filter 100 000 calls per second from a blacklist of 100 000 entries. Blacklist entries could have been any combination of regex-like wildcard characters.

Our final solution exceeded the performance expectations of the client. The system successfully reached a matching of more than 1 million calls per second in a blacklist of any size (more than 100 000)