UST035

51 We spoke with Soliton Systems about the underlying technologies of its new ZAO-SH flagship product, an H.265 video encoder developed to provide live streaming of high-resolution camera data at glass-to-glass latencies below 40 ms. Underpinning the encoder is Soliton’s proprietary RASCOW algorithm, in its newest iteration, RASCOW2. This has been developed using the company’s multi-link bonding technology to fuse the networks of different cellular carriers together to provide reliable, secure and redundant live video transmission amid the constant fluctuations in speed and latency typical of mobile networks. As Kerim Sare explained, “RASCOW is an RTP-based streaming protocol optimised for wireless networks, and can use multiple networks thanks to its bonding and load-balancing technologies. It checks the network and controls the encoder for adapting the transmission settings to continuously changing network conditions in order to provide stable video transmission.” The company’s algorithm also has packet-sort and error-correction technologies for reliability and authentication, and AES 265-bit encryption for secure point-to-point live video transmission. Sare added that the RASCOW2 algorithm includes parallel processing functions and network-handling capabilities to achieve real-time latencies as short as 40 ms, and that these capabilities will be increasingly vital as streaming and control links over 5G networks start becoming commercially available. Soliton Systems has started field tests of the technology using remotely operated UAVs for BVLOS control projects. “In the case of using an H.264 encoder, you will need a minimum of 4 Mbit/s bandwidth to transmit full HD live video data; but with an H.265 encoder, you’ll only need about 2 Mbit/s,” Sare said. “The latter consumes less than 1 Gbyte of data per hour, resulting in significant data cost savings for UAV operators without compromising on video resolution. “The ZAO-SH and our other encoders can be used with public networks as well as private networks. We have tested our encoders with 3G, 4G, LTE, Wimax, Private LTE, wi-fi, satellite and 5G networks; our algorithm can bond all these connections.” Inertial Labs was available to discuss two new product launches for inertial sensing and navigation. The first is the Kernel-100 inertial measurement and digital tilt sensor, which contains three-axis MEMS accelerometers and three-axis MEMS gyroscopes for measuring linear accelerations and angular rates. In addition to extensive calibration and thermal compensation, it has also been mathematically aligned to an orthogonal coordinate system, resulting in 2 º /hour bias in-run stability from its gyroscopes and 10 µg bias in-run stability from its accelerometers. “We’ve designed it particularly with manufacturability in mind,” said Jamie Marraccini. “With these kinds of sensors, manufacturers have to keep in mind the structure of the electronics inside, but during the calibration process, technicians touch those electronics – a lot. For every touch, the cost goes up, and the ability to cost-effectively produce something – as in its manufacturability – becomes less scalable. “In addition to accelerating and automating the Kernel-100’s assembly, we’ve aimed to reduce the number of human interactions needed after that, even for key steps like moving the IMUs onto rate tables and into thermal test chambers, incorporating a great many of the lessons learned over the past 19 years on lowering our products’ price- performance ratio.” The system has also been heavily SWaP-optimised to enable ease of integration across different vehicles and sensor bays. It weighs 7 g, measures 26.5 x 19.5 x 8.5 mm, and typically consumes 0.365 W from a 5 V supply. The company also recently unveiled its new INS-DU, designed as a low-cost navigation solution combining MEMS inertial sensing with a u-blox GNSS receiver and dual antenna inputs (hence the ‘DU’) for calculating and outputting heading information, along with position, velocity and absolute orientation. “Our housings have been designed to be used modularly for a while, and we can support whatever GNSS receivers, IMUs or other aiding sensors – such as airspeed sensors or optical flow sensors for example – that customers seem to be after. The INS-DU is an example of that,” Marraccini added. “It’s built from a mix of low-cost components with high-level aiding AUVSI Xponential virtual 2020 | Report Inertial Labs’ Kernal-100 inertial sensor Unmanned Systems Technology | December/January 2021 Soliton Systems’ ZAO-SH video encoder