Thread Links Date Links
Thread Prev Thread Next Thread Index Date Prev Date Next Date Index

Re: 802.21


Let me say at the outset that you are most welcome to participate in the working
of 802.21. IEEE operates on membership at an individual level rather than
company based. Every participant is a member of the WG, though there are very
specific IEEE attendance rules to acquire voting membership.

At this stage of 802.21 working, the emphasis would be on participation and
constructive ideas to solve the problem of Seamless Media Independent Handover.

Regarding your concern about the complexity of cellular systems as opposed to
802.xx systems, I am not so sure. The main difference between the 802.xx systems
and various cellular systems is that 802.xx primarily defines the MAC/PHY layer
specification, whereas, cellular systems not only define the MAC/PHY of the
air-interface but the whole network architecture as well, end-to-end.

But at the same time, if one tries to take a larger view, we in 802.21 are not
talking about going into the inner workings of cellular MAC/PHY but stay above
that. For example, one possible idea that has gained traction is Layer 1/2
triggers, which could facilitate timely/informed handover decisions. Obviously,
a handover decision would still have to be made by some higher level entity.

Again, issues related to Authentication, Authorization, and Accounting (AAA) as
well as QoS under different air-interfaces and networks would play a role. My
personal feeling is that some of these issues are more implementation related
and others would have to be dealt with at a policy level.

Besides the above, the impact of your point regarding "legal interception" would
have to be understood more clearly. I may just add that here the emphasis is on
fast handover (make-before-break wherever overlap exists) otherwise protocols
such as Mobile IP perform fairly well in heterogeneous networks.

I would like to cc this email to the larger group through the 802.21 reflector,
so that others in the group can also participate in the discussion.

Let me know if you plan to give a presentation so that I can schedule you on the

Chair, IEEE 802.21

On 4/14/2004 9:14 AM, Reijo Salminen wrote:
> Ajay, David
> Since the next 802.21 meeting is not too far away in the future, and in case
> I will be allowed to present my ideas in the meeting (and hopefully be
> considered as qualified member to participate in the work), I would need an
> indication for that, since to prepare a proper presentation would take some
> time for me and is as you know quite an effort anyway.
> Some words about my background - I've been involved in the development of
> different cellular systems - NMT, GSM (from the beginning), PDC (aka
> Japanese Digital - in HLR and MSC), WCDMA (in core network level, MSC
> server, MGw and in RAN level and RNC L2 user plane), all these in different
> roles from design to product management, and before the cellular systems I
> have been working on transmission systems and speech processing, and have
> done research on digital video systems. These all in different companies
> here in Europe.
> What I tried to say in my previous mails need a bit more clarification, and
> I try to explain a bit more here.
> I understand that it is a very natural thing to try to integrate different
> communication systems (even wired, but mainly wireless) by using handovers
> between the systems, and as you have stated it is not too difficult to be
> done in the IP and 802.xx 'world' systems. But when the cellular systems
> come into the picture the situation will be somewhat different, I'm afraid.
> If you look at the TS'es of 3GPP you get an understanding what I mean. You
> have also raised in your meetings the question about how to handle security
> and authentication in connection to the handovers between arbitrary networks
> - that is definitely a challenge, as will be the legal interception that I
> mentioned in my earlier mail. Also it should be kept in mind that increased
> complexity will always mean lower in-service performance and higher cost of
> ownership of the systems, this is a lesson that has been learned the hard
> way in the cellular world. Also it can be difficult to find the needed level
> of expertise to be able to handle the technical detailed questions that
> arise when the mobile host is performing handovers between different systems
> - and who will then take the responsibility if/when something goes wrong -
> the consequences can be surprisingly big in the worst case.
> What I'd propose to consider is to see the different networks (for example
> 802.11x, 802.16, 802.20, UWB, GSM, GPRS, EDGE, CDMA, WCDMA, CDMA-2000 etc)
> as 'components' and build additional functionality to combine the needed
> 'components' into the heterogeneous system so that the smallest amount of
> interworking impacts are caused to the individual 'components'. This way it
> could also be possible to use the different technologies in the way they
> have been intended to be used and are optimized for. Also the cellular
> systems have quite well working mobility management functionality which
> could be useful for establishing the connections also in the heterogeneous
> network. So what I mean is that it could be possible to let the mobile host
> have the possibility to use any of the component networks (several of them
> simultaneously if needed) if the heterogeneous network system is built so
> that the needed additional (which is not very complex) support for that in
> place. This would mean that the mobile host has the controlling role in the
> traffic handling, once the network has authorized the mobile host, this is
> somewhat in line if we look for example how the mobile terminals have grown
> in functionality in the cellular system generations versus what
> functionality is in the network nodes. The common denominator to be used for
> decisions on which of the components the mobile host would be using is then
> based on the measured QoS (note I do not mean end-to-end QoS here), since
> that indirectly includes very many different characteristics of the radio
> links of the different access technologies. So at connection establishment
> the QoS of the connection is required from the access network, and if it can
> not be fulfilled, the mobile host initiates establishment of another
> connection on another access network, and when that is set up, the
> application using the connection is informed about the situation and it can
> then perform the 'handover' (or maybe better word in this context is 'swap')
> in the best suitable way, and the old connection can be released.
> There is a lot of details that would need to be explained by using drawings
> and sequence diagrams, but hopefully you get the big picture from this - if
> the 802.21 is done in this way then each of the 'component' networks can be
> handled separately and the security etc questions can be handled in more
> manageable way.
> Comments/Questions?
> Best Regards, Reijo Salminen
> -----Original Message-----
> From: Ajay Rajkumar []
> Sent: 29. maaliskuuta 2004 17:16
> To: Reijo Salminen
> Cc: 'Johnston, Dj'; Ajay Rajkumar
> Subject: Re: 802.21
> Reijo,
> I have been traveling ever since I came back from IEEE meeting in Orlando. I
> am
> currently in UK, but would be back in my office by the end of the week. Will
> respond to your email then, if you do not mind.
> -ajay