The problem of piracy has existed for decades, but identifying illegal behavior has become far more challenging as consumers continue switching to OTT video options.
Streamed pirated content typically looks exactly like a typical, legal OTT video. And while you could use deep packet inspection to identify and track pirated streams, it is cost prohibitive and raises privacy concerns. Machine learning offers a less intrusive and cheaper alternative. By training it to detect variances in meta-data, byte distribution, packet lengths and inter-packet time, AI can identify the illegal streams with a high level of accuracy and precision. NCTA vp, broadband technology Matt Tooley also believes AI can be used to go out and find the video pirates themselves.
“The video pirates have only three or four backend services that they will stream from,” Tooley told attendees at an SCTE-ISBE Cable-Tec Expo panel on piracy Wednesday. “We find those four backends and we can find it all.” But once pirates notice that cable operators have found a way to identify illegal streams, that very well could change. There is the potential that they could use packet stuffing to make their flows a constant bitrate, but they won’t make those sorts of investments without some poking and prodding. “The pirates are cheap,” Tooley said. “They don’t encrypt any of their video streams because it costs money.”
He’s also working on fine-tuning how to ensure AI’s effectiveness for larger MSOs. That means continuing to work with the AI, teaching it to ignore activities like music streaming or online gaming sites that also have video features. That will lower the amount of false positives, ensuring the AI’s accuracy. “There’s a bit of work to do there so we can make this scale to a cloud service so that when the operators want to use it, it makes it easy for us to do it for our members,” Tooley said.
Of equal concern to cable operators and beyond are the DDoS attacks that have the possibility to cause serious disruptions on their networks. Buying an attack against a company is relatively cheap, costing the buyer only $20/hour for 100Gbps or more. Victims forced to fight back face a much higher price tag, spending on average $40K/hour to put an end to the attack. The risk landscape has only expanded as more people adopt IoT devices and place them on their network with little to no knowledge of how that device is actually interacting with that network.
In his work in attempting to detect vulnerabilities in networks and when a DDoS attack could occur, CableLabs senior security engineer Kyle Haefner found that a device as simple as wireless tea kettle could prove usable by hackers. Driving the point home was a slide in his presentation dubbing the iKettle from UK-based connected home company Smarter the “WiFi enabled tea kettle of doom.” “They can tap into your iKettle… and it gives the attacker the unencrypted WiFi password of your wireless network,” Haefner said. “You have to be a network manager in your own home, and most people don’t have that skill.”
CableLabs is working with the Messaging, Malware, Mobile Anti-Abuse Working Group (M3AAWG) to collect and aggregate attack data through the DDoS Information Sharing Project. The pilot began in 2017 in an effort to provide ISPs with data they can act upon to rehabilitate compromised devices on the networks before they become compromised.
Read more here:: feeds.feedburner.com/cable360/ct/operations?format=xmlPosted on: October 3, 2019