Thursday, July 29, 2010

1300-1500 Afternoon Session I

1. Introduction, Rajeev Koodli (Chair), 5 minutes

Note taker – Ashutosh Dutta

Rajeev gave a status of MOBOPTS RG and also solicited ideas for future IETF MOBOPTS meetings.

4 RFCs have been published so far and one draft has completed IRSG review.

Media Independent Pre-authentication Framework Document IRSG review concluded, authors will update the document based on the last set of comments received.

2. Inter-access mobility with multiple protocols - Carlos Bernardos,

20 minutes

Carlos presented performance results of several mobility protocols. Analysis of the combination of different IP mobility schemes

Ubiquitous mobility

Mobile IPv6, Proxy Mobile IPv6 were taken into consideration.

Extension to PMIPv6 to support mobile MAGs

Different combinations were considered

1. MIPv6 and PMIPv6

2. NEMO B.S +PMIPv6 +MIPv6

3. MIPv6 + N-PMIPv6

4. NEMO B.S + N-PMIPv6

5. NEMO B.S + N-PMIPv6 + MIPv6

Performance analysis:

Pros: Overhead and Handover latency

Cons: L2 handover

Choi – Solicited Router advertisement is used, why not DNA?

Answer: It is dumb node, it waits for router advertisement (RA) instead of sending router solicitation (RS)

Thomas: Router advertisement variance

Hakim – Re-configurability, IP addresses

Question - User preference on the mobile client. More than overhead problem on the node

Thomas: Handovers for different Mobile IP protocols

Ashutosh: It would be good to compare the performance of these protocols in a real experimental testbed.

3 different scenarios were discussed and their performance figures were presented.

Local: 5.37 ms

Regional: 18.32 ms

Continental - ?

3. 4G Mobility Management -- what went wrong? Charles Perkins, 30

minutes

Charlie described what may have gone wrong in 4G mobility management, usage of Mobile IP and GTP etc.

Why Mobile IP did not play any role – Improved Mobility LTE Management

LTE will predominate in WiMAX and Charlie pointed out the role of Mobile IP.

Charlie first gave an overview of LTE. There were questions about how many tunnels can a PDN gateway sustain for scalability.

Rajeev – PDN gateway can sustain few million subscribers

Constraints and Goals

Lei – Requirements of LTE mobility, overcome lack of mobility within LTE

Circuit switch fall back is an option.

Rajeev: There are different choices, if there are different. How to do Voice over LTE. Offered as service

AVI: Approach taken by 3GPP2, IMS

Given the signaling level. IMS is heavyweight

Question: IP from the mobile node, cellular network. No IP number associated with the IP-Phone.

Charles: Handle the handover, doing Mobile IP between the LTE and Non-LTE networks or within LTE.

S-GW Issues: Handover between other networks

Dave (Verizon) – P-gateway assigns /64 address

If you move between P-GW, then IP address will be changed. If you use Mobile IP to take care of.

Rajeev: One could still anchor it on the P-Gateway. How much routing it enters in the Internet.

David: Option to hand out static address or /64 address

Keong (ETRI) – Policy issue: Separate S-GW and P-GW

Avi – Spec does not require S-GW and P-GW to co-locate

Charlie mentioned that past projects have exhibited excellent performance using Mobile IP

S-GW+P=GW = HA-D

pMME = HA-C

Aggregation of mobility events

P-GA and S-GW are tightly bound

Fully routed EPC seems likely to work better

Shown a new picture by introducing

Avi- Enode B can change every 5 seconds

Yokota- Does it belong to 24.01 or 24.02– combined functionality is already defined in the document

Tunnel between HA-D and S-GW

Split HA into HA-D (==P-GW + HA-C (==MME)

Enable HA to tunnel via GTP

New NAS attach request message

UE pMME

4. GENI Update, Aaron Falk, 30 minutes

Aaron Falk from BBN – GENI Engineering Architect provided

GENI status.

He had given some GENI related talks earlier also.

How the Internet interworks with human system.

Conceptual design – three fundamental pieces to it. Change the behavior of the systems make the systems programmable and virtualized

Heterogeneity – Join interesting pieces

How we will use GENI:

Analysis, simulation, use the Internet to get the experiments

GENI Spiral development: Nearing the end of second spiral of GENI

PLANET Lab Europe: How we get these experimental facilities to IETF.

It would be interesting to have test-beds.

Aaron: IRTF might be interested in using GENI Testbed

Rajeev: IETF does some prototype experiment. NAT64 is being tested out. IETF organizers might be interested to try out some of the GENI related experiments.

Aaron then discussed the status of some of the existing GENI projects.

ViSE, Wash U, ORBIT, Rutgers WINLAN, DieselNet, WAIL, U Wisconsin-Madison, DRAGON

LambdaRail

OpenFlow – Stanford, U Washington, Wisconsin U, Indiana U

ShadowNet – based on Juniper Networks

WiMAX – It can allow multiple subsets of traffic, programmable base stations.

OpenFlow evaluated in the campus

Controllable experiments

Solicitation out – August 20, 2010

Build out the architecture

Experiment support/training/education and curriculum

Virtualization in the router

Washington University in St. Louis has programmable router.

OpenFlow switches

G-LAB. FIRE, Brazil, ETRI, JGN2plus, NICTA

GENI workshop is scheduled for November 2-4 2010 and is open to all.

6. G-Lab, Thomas Schmidt, 20 minutes

Thomas gave an overview of research activities from G Lab. It is a National Initiative. It is trying to do two things: 1) Heterogeneous collection of projects. 2) Bringing work to IETF. Because of lack of time Thomas could not spend all 20 minutes and needed to shorten his talk. He described different phases of the projects in G Lab.

Phase I, Phase II

Phase I – Key component of

Proposals of FIR Architectures

Hybrid Mobile Multicast

Global scale of

Large cluster – 4 Control node

Partner location

Low delays

Planet Lab Delay space

Jitter pictures

G Lab is open to extension

7. Formal Approach to Mobility Modeling, Ashutosh Dutta, 15 minutes

Because of lack of time Ashutosh could talk only for a limited time. In his brief presentation Ashutosh discussed a formal approach to handoff management that could be useful for mobility deployment. Ashutosh pointed out the common abstract functions among all the cellular and IP-based mobility protocols that are needed during handoff, how dependencies and trade-off between performance and resources (e.g., energy) should be taken into account during any handoff design and elucidated the need for a mobility model that can choose the right types of mobility protocols or functions during deployment. Since there are many mobility protocols out there, he described the need for a best common practice document that would help deploy the right type of mobility protocol for the right environment. Many in the audience including Charles Perkins liked idea of having a common best practice document. Ashutosh proposed to the chair that he plans to submit a document describing the best current practice along the line of his presentation.