If a network drops a packet or experiences a latency hiccup, most of the current crop of consequences are bearable: a video stutter, res-downgrade or buffer-swirl on Netflix; ‘some text missing’ in a standard SMS message; or an undeserved frag in a multiplayer shoot-out.
In the realms of remote surgery, events of this nature really can signal ‘game over’, particularly if an anomalous – rather than dropped – packet quite literally sends the wrong signal momentarily to a robot that’s performing a millimetre-critical telesurgical procedure. Data glitches during cybernetic coitus are likely to be less injurious, but to just as emphatically kill the mood; and at the very least, poor latency in biofeedback is likely to cause the same kind of ‘cyber-sickness’ that gamers can experience when the ‘equal and opposite’ reaction they were expecting wipes its feet at the door.
A group of researchers from the Institute of Electrical and Electronics Engineers (IEEE) are considering [PDF] these and other impediments to the development of the ‘haptic internet’, a touch/pressure-based iteration of the internet which, they believe, will ‘revolutionise almost every segment of society’ – if a massive leap of network quality can be achieved.
To this end the researchers propose changes both in the way that haptic information is transmitted and received, and in exploiting the multiplexing capabilities of 5G to bring near-‘real-time’ feedback without the high overhead of a TCP approach or the unreliability of a system based on User Datagram Protocol (UDP).
Achieving the gulp-inducing 1ms latency required for long-term and highly critical network haptic procedures presents a number of challenges, some of which are ironically being caused by network improvements. The extra overhead of IPv6 potentially doubles the transmission time for haptic information, turning the need for a codec-based approach to pressure/tactile data transmission from a luxury to a necessity.
‘A fundamental challenge in context of the Tactile Internet is the development of a standard haptic codecs family, similar to the state-of-the-art audio (ITU-T H.264) and video (ISO/IEC MPEG-4) codecs. Embracing both kinesthetic as well as tactile information, such a codec family would be a key enabler for scalability at the network edge and universal uptake. Besides, it introduces a layered approach to haptic data (comprising multi-modal sensory information), which would be crucial for operation in typically challenging wireless environments.’
The research acknowledges that the CAPEX and OPEX factors in considering new haptic-specific infrastructure are prohibitive, but that the versatility of 5G networks would make feasible a reliable haptic network by employing Network Function Virtualisation (NFV) and Software Defined Networking (SDN), and the IEEE researchers assert that the ‘network function can be managed as a software module that can be deployed in any standard cloud computing infrastructure. On the other hand, SDN provides an architectural framework wherein control and data planes are decoupled, and enables direct programmability of network control through software-based controllers.’
Virtualising the radio spectrum of 5G in this way permits a slice-based approach that could potentially take haptic information protocols to carrier-grade standard – a vanguard effort that’s necessitated by the critical nature of the problem, but which seems likely to have a beneficial ripple effect on other network operations if successfully adopted. The multiplexing codecs envisaged by the research – and prior papers – take in the possibility of combining several necessary data streams – including standard streams such as video and sound – along with the haptic push/pull (etc) information. And none of the other current commercial imperatives in the marketplace seem as likely to drive change as the requisites for transmitting this particular type of information.