Huawei Set for 'Intensive' AI Investment

LONDON -- Global Mobile Broadband Forum 2017 – In its efforts to provide telcos with the "intelligent" systems they will need to build fully automated networks, Huawei is set to make "intensive investments in AI" in the coming years, according to the vendor's deputy chairman and rotating CEO, Ken Hu.

Huawei Technologies Co. Ltd. is making a big push on helping operators improve not just network performance (speed, latency, capacity) but also network intelligence, so they can manage the evolution to 5G and take advantage of new business opportunities, such as those related to IoT. Future networks, noted Hu, need to be self-optimized, self-healing and automated, all of which needs AI capabilities, reckons the Huawei chief. "Full network automation is what we are aiming for and we've been working on this for a few years," noted Hu.

So just how much is Huawei going to invest? Hu would not be drawn on an exact figure when pressed by Light Reading, though the company is noted for putting significant funds behind its major plans -- it invested more than US$11 billion in R&D last year, for example.

Huawei isn't just starting on its AI work, though. Talking to media and industry analysts here today at the vendor's annual mobile broadband event, Hu said AI capabilities (machine learning) have already been introduced into some mobile network management and fault management systems.

Hu's colleague, wireless networks CMO Peter Zhou, noted that SoftBank Mobile Corp. in Japan is already using AI-enabled technology from Huawei to automatically adjust MIMO parameters in its advanced 4G networks, while Vodafone Turkey is using machine learning and advanced diagnostics to drastically improve fault management in its VoLTE infrastructure.

That, though, is just the beginning. Huawei has a number of AI proofs of concept (PoCs) underway across different parts of the Vodafone Group plc (NYSE: VOD) -- LTE traffic prediction and VoLTE packet loss root cause analysis in Germany, enhanced carrier aggregation and load balancing in Spain -- with further trials and commercial deployments planned in 2018 with Vodafone.

Hu regards AI as a general-purpose technology that will be widely adopted, "just like electricity today," with operators expected to widely deploy machine learning capabilities to "learn from the large volumes of data" already available to them and, as a result, improve efficiencies. To take advantage of AI, telcos will need compute capacity and data sets, noted Hu, who added that the latter "is not a problem for the telcos -- they have a goldmine of data. AI [and analytics] will enable them to dig deeper into that goldmine."

The CEO believes that AI will be embedded "end to end in all telco processes" within ten years, and that the planned "intensive investments" will help Huawei develop "core competencies" that will complement industry collaborations (though he didn't identify any specific partners).

Hu also noted that AI capabilities will exist not only in the cloud but also in end-user terminals -- Huawei has already integrated a neural processor into its latest Mate 10 smartphone. (See Huawei goes all-in on AI with Mate 10 smartphones.)

So what will the broad deployment of AI and extensive automation mean for the telcos? Efficiencies are already starting to eat into operator headcount and the expectation is that this trend will only continue. (See Efficiency Drive by Major Telcos Has Claimed 74K Jobs Since 2015 and The Revolution Will Be Automated .)

Hu wouldn't offer any details, saying only that the impact would be "profound," and that there were a lot of benefits to be gained in terms of operational efficiency, especially as, according to Huawei, 70% of network faults are caused by human intervention, something that full automation would rectify.

Huawei is not a company that does things in a small way and with three major business divisions -- carrier, enterprise and consumer device -- that can all make use of any internal AI investments, it seems likely that the "intensive investments" in AI are likely to have a significant impact on the vendor's near-, medium- and long-term business opportunities.

— Ray Le Maistre, International Group Editor, Light Reading

Be the first to post a comment regarding this story.
Sign In