clusterEreviewmeeting: Cluster-E-meeting-notes-29Jun09.txt

File Cluster-E-meeting-notes-29Jun09.txt, 8.3 KB (added by hmussman@bbn.com, 12 years ago)
Line 
1Cluster E Meeting Notes
2June 29, 2009, 1pm -4pm
3GPO: Chip, Aaron, Heidi
4Rutgers: Ivan, Ray, Meiling Zhang (NEC)
5
6Goals: review progress, GPO suggested plans, project suggested plans
7
8Agenda:
9  - WiMax Project
10    - project review & plans
11    - GPO milestone review
12    - begin roadmap discussion
13  - Cognitive radio platform status update
14  - GENI goals
15  - Cluster E roadmap topics and discussion
16  - OMF project review
17    - project review & plans
18    - GPO milestone review
19    - continue roadmap discussion
20
21=======
22
23WiMax project slides - Ivan
24
25System:
26  Chassis includes control module and 3 radio "channel" cards
27  Outdoor unit with RF subsystem
28  PC controller running NEC software and researcher software
29    plan is to merge chassis and ODU
30  Fully WiMax compliant
31
321st goal is to demonstrate proof-of-concept: done
33
34Cards:
35  Works with Intel PC cards (need to make base station pretend to be
36  Sprint)
37    The right way is to create a provisioning server to tell cards to
38      accept 'GENI' base stations
39  No handsets in US, maybe Korea or China
40
41Installation:
42  Single antenna installed at WINLAB tech center 1Q09, 'illegally'
43  mounted   on building, needs to be certified by a Professional
44  Engineer; could be a cost issue for other deployments
45
46  Research license requires deployment to be less than 6ft above roof
47
48  FCC license awarded for two sites (rt 1 and campus) quickly (~4wks),
49  checked BBN and NSF references
50  Constrained to use within 10km
51
52  Actual coverage 5-10km
53
54Controller:
55
56  PC controller, "ASN" - access service network; planning to keep
57  NEC-provided controller in place, will expose as many interfaces as
58  possible; Rutgers NEC group has source will expose any and all
59  capabilities necessary;
60
61  WINLAB writing software for controller handling control plane
62  functions, scheduler shaper, and TUN device that handles mapping
63  between clients and slices.
64
65  Currently working on virtualization; using a Click router to handle
66  virtualization; want to move beyond that to using a custom virtual
67  driver that handles mapping of slices to queue performance
68
69  Still moving towards supporting slices.
70
71Spiral 1 Milestones
72
73  Base station API and L2 controls - done
74
75  Virtualization: need to do click-based router, integration &
76  evaluation
77
78  Full OMF integration: ...
79
80  Demo will be a little ahead of what was promised
81
82  Slice-based algorithms require some research...
83
84Spiral 2 goals (Ivan)
85  - L2 connectivity from Rutgers to PHL PoP
86  - Run R6 interface over GENI backbone
87  - measurements
88  - improve operator mgmt GUI
89  - better documentation
90
91GPO: Spiral 1 status (Harry)
92
93  - Should produce a top-down diagram of network integration; problems
94    getting from MAGPI to Internet2
95
96  - Collaborations with O&M: no contact yet; project has XML data
97    export that is documented; Ivan will send to Heidi/Harry for
98    forward to GMOC
99
100  - Collaborations with security: no progress yet; action to call
101    Schwab and explain project approach to access control
102
103  - Outreach: Ivan doing tutorial on FI school, 'huge' summer
104    internships
105
106Cognitive Radio (Ivan)
107
108  Orbit includes without software control plane: USRP, Rice WARP
109  Platform, U. Colorado, USRP2, RST SDR system, WINLAB WINC2R system
110
111  Goal: look at fairly large scale deployment of cognitive radios; no
112  one is looking at this now and that is where the benefits are;
113
114  Very active in spectrum allocation issues; seeing more take up these
115  issues in industry
116
117  Unique design goal is to have four radio channels that can be
118  independently managed, say, 1 for spectrum sending, two
119  transmitters, and 1 receiver
120
121  Want to be able to do experiments currently possible with ORBIT
122  using more than 50 cognitive radios
123
124  Toyota funding work in spectrum sensing in car-to-car networks; RF
125  and AtoD are more important than folks appreciate
126
127GPO's Spiral 2 Goals (Chip)
128
129  Support for continuous experiments
130
131  Identity management - possibly using InCommon;
132
133    Ivan: should be straightforward to add new identity technologies;
134
135    Ray: easy for different projects to pick different approaches to
136    all these topics, GPO should take a stronger hand to get better
137    interoperability
138
139  Improved integration: maybe better with NICTA, OneLab
140
141    Ray: wireless is often on the lower side of the integration
142    hierarchy, requires someone to take responsibility of the wired
143    piece
144
145  Instrumentation: should have lots of instrumented things and
146  beginnings of how to archive and find measurements; need some good
147  ideas around this
148
149    Ivan: Driver should be a person who has a keen interest in
150    processing accumulated data.  Lots of data being dumped.
151
152    Ray: two types of researchers in GENI: 1 that sees the Internet as
153    an uncontrolled system to be observed and the other that is
154    interested in controlled and repeatable experiments
155
156GPO: Spiral 2 Roadmap Goals (Harry)
157
158
159  2.1 a - Experiments & users: now working on coming up on a set of
160  tutorials;
161
162    Suggest organizing something like the EU's 5-day Future Internet
163    summer school, pay folks to give tutorials, bring in students,
164    have regionally, limit to 200 miles driving with hands-on project
165    workshop
166
167  2.1 b - when can WiMax testbed be used?
168
169  2.1 c - Making WiMAX available to GENI users earlier than planned
170
171  2.1 d - Intel WiMAX cards?  Maybe register GENI as a carrier?
172
173  2.1 e - accessing servers?
174
175    Ray: we have a plan for accessing PlanetLab nodes.
176
177  2.2 - Identity management
178
179    Ivan: willing to integrate with any framework's identity
180    framework, prefer to pick only one though
181
182  2.3 Interoperability
183
184  2.4 Instrumentation
185
186  2.5 Interoperability
187
188    Already working with Enterprise GENI to let OMF work with a
189    OpenFlow controller.
190
191    Ray: suggest creating a shim or pick the best and force the others
192    down.  Will reduce the combinatorial challenge.
193
194=========
195
196OMF Project Review & Future Work Plans
197Max Ott
198
199  Reviewing project SoW & achievements (see slides):
200
201  Extend OMF to support multiple heterogeneous testbeds
202    - Deployed OMF beyond ORBIT
203    - Refactoring components and aligning names with GENI
204  Disconnection tolerant framework
205  GEC4 Demo: Disconnected demo
206  Provide VLAN connectivity via NLR & MAGPI East
207  Establish OMF operating environment for testbeds, verify with
208    experiments, & demo
209
210OMF Future Work (Max)
211  - Federation
212  - Resource Description
213    - want a formal model that supports multiple levels of abstraction 
214  - Everything needs to be observable
215    - measurements are crucial
216  - Policy & Authorization Framework
217
218Also working on:
219  - Onelab2 (EU, FIRE)
220    - includes PlanetLab integration via geni-wrapper
221  - 4WARD (EU)
222  - NaDa (EU, FIRE)
223  - Continuous Bridge Monitoring (Australia)
224    - using Motes, UMTS
225
226Federation
227  - using a pub-sub, message-oriented architecture based on XMPP
228  - need distributed policies, use of SAML & XACL
229
230Year 2 plans:
231  - Extend OMF to support experiments across multiple testbeds
232  - Extend OMF to support experiments driven by context, such as
233    location
234  - Continue working with WiMAX base station
235  - Develop a resource description model for our resources (based on
236    NDL/OWL/G.800)
237
238  Re: shibboleth - no pushback from OMF, should work, better if others
239  use it too
240
241Controlling mobile experiments with spatial and temporal triggers,
242(Marco)
243  - define markers on a Google Earth map, trigger experiment actions
244    when triggers are hit
245  - e.g., using ultrasonic sensors in cars to sense whether spaces are
246    vacant
247  - another project doing speed estimation
248
249Year 2 stretch goals
250  - experimenting with user mobility
251  - better integration of OpenFlow
252  - support teaching labs based on testbed experiments
253
254GPO Spiral 2 Goals (Harry)
255
256  - New experiments: mobile parking sensing network; Chip: would like
257    to see outside teams and more diverse infrastructure
258
259Meiling Zhang (NEC) update
260  - NEC Japan very interested in GENI
261  - Working to make base station software easier to obtain by
262    researchers
263  - Would like to see more roadmap for Spiral 2
264  - Should be able to add virtualization and authentication/AAA in
265    WiMAX; if NEC Japan sees more progress, they're more likely to
266    increase commitment
267