Thursday, February 3, 2011

Eduroam the key to a National Public "IPv6 only" R&E Internet

[As many have you seen in the press there has been a lot of discussion on overloaded 3G/4G networks and the need for WiFi handoff to handle the huge data volumes now flooding 3G/4G networks
due to the explosion of smart phones and pads.  Many telcos and cablecos are deploying WiFi nodes in data hot spots to relieve the traffic congestion on their 3G/4G networks.  Of course, one of the most common hot spots for carriers is university campuses as students and researchers tend to be heavy users of wireless data.  Although students and researchers nominally should have access to campus WiFi data network many do not configure their smart phones and pads for local access.  Also most visitors to a campus, the overwhelming majority who are not from another university and therefore do not have Eduroam, place significant loads on 3G/4G coverage at universities.  In addition to the traffic load on campuses university, researchers now are demanding access to a variety of widely distributed sensor networks through their 3G/4G networks. Researchers are also starting to deploy applications such as personal health monitors, automobile traffic analysis, etc that use the smart phone as a mobile sensor.

For these many reasons I have long advocated that universities and R&E networks need to explore deploying national wireless R&E networks that complement their existing fiber infrastructure. Because of the huge load of 3G/4G data traffic on campuses I believe there is an opportunity to negotiate with some 3G/4G operators an exchange of services.  3G/4G providers can be provided access to campus network in exchange for university and/or national R&E network having access to the 3G/4G provider’s network including WiFi hot spots.  There are a number of possible arrangements, but I strongly support Rudolph van der Berg’s  (http://internetthought.blogspot.com/) idea that, as a minimum, a national R&E network should be a MVNO (Mobile Virtual Network Operator)  with its own IMSI number (International Mobile Subscriber Identity).  With this arrangement a national R&E wireless network can negotiate roaming agreements with other national wireless R&E networks, and if necessary switch 3G/4G providers or have multiple providers for the 3G/4G service that underpins the MVNO.  This allow the R&E network to provide continuous service to the many upcoming fixed and mobile sensor networks being deployed by researchers.  With the new software SIMs becoming available in next generation smart phones, pads and PCs, campus IT folk or national R&E network operators can program these SIMs remotely so that students and researchers can access the national R&E MVNO as well as the services from their commercial 3G/4G provider. This also be useful in emergency situations as it will be far easier to access all student phones in moments of a crisis.

A number of companies are deploying integrated WiFi/3G/4G pico cell devices that could easily be deployed across the campus. The Bel-Air WiFi pico cell product is a good example (http://goo.gl/ogjRo).   Some of these devices can be managed simultaneously by the university IT folk as part of their campus WiFi service as well as the national R&E MVNO and/or commercial operator providing 3G/4G services.  Another option is to deploy open source 3G base stations such as OpenBTS (http://goo.gl/T2Smo). Of course many of these devices can easily be powered by solar panels or micro windmills using PoE Power over Ethernet or 400 HZ multiplexed power systems.  The advantage of PoE or 400 HZ is that the solar panel or micro windmill can be located some distance away from the actual Wifi/Pico cell node.  And as I mentioned before in my 5G wireless network blogs (http://goo.gl/CAJcu) there is no need for backup power as multiple nodes can provide overlapping coverage, and in the worst case default coverage can be provided by 3G/4G service from an off campus.  In some situations service providers are installing these WiFi/pico cells at no cost to the university – however I would be very cautious about this arrangement and make sure you negotiate a suitable quid pro quo (such as a national R&E MVNO) as well as shared management of the device.

One of the challenges of integrating Wifi hot spots with 3G/4G networks is the authentication and authorization handoff between the 2 networks.  A number of organizations are working on gateway protocols to allow for this seamless handoff, but they are very complex and so far unproven. The cell phone companies are working on something called WISPr 2.0 (http://goo.gl/poYRf) which as you would expect from the telephone company is inordinately complex and convoluted. The Wifi alliance(http://www.wi-fi.org/organization.php)  is developing a competitive standard called Hotspot 2.0 built around some of the new IEEE WiFi standards.

However, the university R&E community already has a well established working standard for Wifi handoff called Eduroam (www.eduroam.org).  Eduroam is unique in that it is a federation of WifI authentication services operated by institutions.  To my mind this is much better scalable enterprise solution  than the network solutions proposed by carriers.  Originally I was not a big fan of Eduroam, as I saw it as tool that effectively restricted access to information at a university. I believe that universities should be open to all members of  the community, not just visiting faculty and staff from another institution.  This is why I have long advocated for open WiFi for all public sector institutions. But off campus, it is a different story.  Tools like Eduroam are essential to provide services off campus as part of a national R&E wireless network.   The nice thing about EAP authentication used in Eduroam is that you can put almost anything inside it such as certificates, SIM authentication etc.. See EAP-SIM (RFC 4186) and EAP-AKA (RFC 4187) 

In addition to arranging to become a MVNO, national R&E networks could also negotiate with numerous hotspot providers  at coffee shops, restaurants, airports etc to extend Eduroam to these facilities as well.  Also many communities are deploying open access networks, and again an exchange of wireless offerings and content peering would allow citizens of these communities greater wireless access and cheaper Internet, and in turn students and researchers would have wireless coverage throughout the community. A community Wifi trial in Japan, using Eduroam is doing exactly this (http://www.eduroam.org/index.php?p=media&id=14).  Next generation FTTh networks such as those deployed in Amsterdam are also exploring the use of devices that provide WiFi from each connected home at the Optical Network Terminal.  See the  NetU device to enable WiFi or open micro WiFi/GSM cells for FTTh networks http://www.dadamotive.com/2010/04/gigabit-society-broadband-as-a-utility.html.

Finally I believe that R&E networks have an important innovation role to help propel society to the next level of interconnectivity.  As I have blogged in the past “What comes after IPv6 and DNS”  (http://goo.gl/DcNvP) we are in a Mexican standoff with regards to IPv6 deployment. These sentiments were echoed by Geoff Huston in his article “Is the Transition to IPv6 a "Market Failure," CircleID  (http://goo.gl/u0IOp).  I don’t think there is sufficient market incentive to move to an IPv6 world and regrettably if don’t do something we will be stuck in world of carrier class NATs , firewalls and a fragmented Internet.   With the ongoing threat of carriers and content providers wanting to do deep packet inspection and block or divert what they consider to be unwarranted traffic  I think we need a national public Internet  that supports the fundamental end-to-end principle and bypasses all these current threats to the existing Internet. To that end, I believe an important community leadership role that could be undertaken by R&E networks is to deploy a national MVNO that carries only native IPv6 traffic. No IPv4. No dual stack.

The failure to move to IPv6 represents a clear market failure and is something that commercial carriers are reluctant to do – and hence justifies government investment and support for such a development.    I also believe such a network should encourage or mandate that all traffic be encrypted and promote content peering to open access community networks.   Providing transit to open access community network is fraught with a variety of commercial and regulatory risks. But providing IPv6 peering content as many R&E networks do now will further encourage and promote the adoption of IPv6.  And if this was done with exchange of Wifi/3G services would also expand the footprint of a national wireless R&E network.

No comments:

Post a Comment