Posted by
Tom Donnelly in
Device awareness is being talked about more and more amongst service providers today, particularly by those offering mobile data. What is it? Similar to location awareness, device awareness is visibility into what devices are on the network, what impact they are having on network resources and at what times of the day. Visibility to this data allows service providers to better understand how devices are impacting the network and offers the opportunity to better manage capex and opex while continuing to maintain a high quality of experience for their subscribers.
As you see below, this example taken from Sandvine’s Network Analytics product shows the total active handsets in a sample network and their corresponding bandwidth usage.
By having total visibility and control over their network, service providers can seize the opportunity to better manage and maximize their networks. For instance, they can consider partnering with device/application providers or offer subscribers more competitive and innovative pricing tiers. The marketing organization can also focus on delivering marketing campaigns that are tailored precisely to the segment that they wish to target. From an operational perspective, engineers can monitor and predict where and when QoE might be affected, due to specific devices and their demand on the network.
One use case that Sandvine has assisted with, using device awareness, is enabling service providers to offer a special tier to those customers wishing to tether their mobile devices. Tethering is the use of your cell phone or other Internet-enabled mobile device as a modem for another device, like a notebook or PDA. It enables you to go online from your laptop, for example, in situations where there’s no other means of Internet access. By being able to identify “tethering” devices, a service provider can offer a service tier or package that caters to the subscribers’ needs.
The case for device awareness will increasingly become a key component of proactive network management, helping service providers to maintain QoE today and plan for future expansions and new service offerings.
Bookmark this article!
Hello
Device awareness is really interesting for us, but can you explain the reliability of this device identification ?
For example, is it linked to the http header, like the User Agent ? If so, I think it may be spoofed.
And if it can be spoofed, then it is difficult to make an offer linked to the device, although global statistics will still be reliable.
Philippe Le Guennec is quite correct in noting the significant complexity and challenges tin achieving accurate device-awareness in a dynamic mobile network environment. Sandvine’s approach is to combine variety of complimentary techniques.
The possibility to forge user-agents is indeed quite real. At present, it is probably limited to the more tech-savvy users, although widespread adoption is only an app away. Additionally, the growing power of modern smartphones adds another dimension to the issue. It is therefore required that Network Policy Control systems have the flexibility to keep pace with rapid technical and market evolution.
Sandvine’s solutions employ four key mechanisms to address this requirement:
1. User-Agent Identification – Although the User-Agent field can be forged, the ability to read this field nevertheless provides valuable information (consider that the majority of apps are based on HTTP).
2. Application Identification and Correlation – Some applications are known to be used exclusively by certain device families (say, “desktops/laptops” vs. “mobile devices”) or operating systems (Android vs. iOS vs. Windows). Consequently, identifying the presence of these applications provides an extremely reliable indicator of specific devices. If a particular application is only “mostly” used by a particular device, then the information is still useful, but you might want to combine it with other potential indicators.
3. Integration to Device Maps – In 3GPP-compliant networks, device-specific information like the International Mobile Equipment Identity (IMEI) code is provided; the ability to obtain this identifier (for instance, from RADIUS traffic or through seamless integration with another network system) without disrupting/reconfiguring the existing network topology is incredibly valuable.
4. Usage Measurement – Measuring traffic volumes and application types over defined intervals is another potentially useful indicator when seeking to identify device types. Although handsets have dramatically increased information appetites over years past, a laptop still provides much greater processing power. This gap in processing power leads to large differences in how network traffic behaves. Compared to smartphones, laptops generally:
• Have much larger aggregate volumes of traffic per subscriber
• Exhibit a much larger number of concurrent applications and protocols per subscriber
• Experience a much larger number of concurrent network connections (flows) per subscriber
Taken alone, any of the techniques above can give a strong indicator of device-type, but most service providers prefer to act on certainties. Taken together, the techniques deliver the certainty on which business decisions can be made and services can be deployed.
Hi Tom,
I wanted to ask several question :
1) Is it possible to block a user from using tethering from PCRF?
2) If yes, what parameters does the PCRF used to decide to block and not to block?
Regards,
Ato
Thanks Ato.
A PCRF cannot reliably detect tethering because to do so would require seeing the bearer plane (data). Accurate detection requires considering many factors beyond the HTTP user agent – for example the number of concurrent flows, application awareness (e.g. windows update), etc. A PCRF accepts signalling via Rx (from an external application function) only, and there is no application function for this. It is however possible to achieve what you ask, and we can discuss in more detail offline if you wish. In general, this is done in the bearer plane in our integrated pcrf/pcef solution.