Subject: RESCUE MEETING
Date: 01/16/04
Loc: URP
Author: AM
Discussion of sharing, data integration, security.
SM: Chen Li may have ways to automatically pull data off Crimemapper.
NSF funding is year-by-year.
One evaluation criterion is:
Advancement of research in our areas.
*NEED PAPERS. Be aggressive re publishing.
Need 15-20 publications in the first year.
AM: [Status]
Found one correlation of CrimeMapper with dispatch logs in RAINS-NET.
- This was tedious, time-consuming to discover.
Calls into Portland PD re Crimemapper data.
Call into Carl Simpson, Portland 911 center.
Need to formulate queries for local agencies, e.g., Irvine PD and others.
At least one North Carolina 911 center wanted a formal request.
SM: [Whiteboard: Sharing – Data Integration]
Mediation – which db has the data, bring it all together.
Organizations sharing data.
Form groups, then overlapping groups.
Design a middleware architecture.
Group-based middleware.
- Must be adaptive.
Heterogeneity of data, systems, policies.
- Differing communication protocols, data formats.
Open, distributed systems for trust negotiation and trust relationships.
Organizational structure
NV: In the presence of real-time failures, extreme conditions.
CB: Trust – preventing malfeasance not as important as trust in capacity, ability to execute tasks.
Overload – too many people trying to help.
Chen? New problems vs building on existing research.
SM: Global integrated data model <= Translation => Local data models.
NV: Schema translation.
Semantics don’t always carry over.
SM: Efficiency in integration, translation.
GIS info – most critical here.
Maps, loop sensor data, etc.
[GIS-data1, GIS-data2, Internet as source-Newsfeeds]
Mediation – too hard in general.
GIS – constrains the problem, maybe doable.
Eg, map resolution may require levels of access, clearance.
- best answer given my credentials and authority.
NIMA, USGS, tax collection maps.
NV: Not just level, but location also (eg, Army base).
CB: Identify semantic space to see similarities.
If we do this, there is enough generality for other problems.
SM: SQL vs spatial.
“Least privilege”.
GS: Least privilege model in computer security – old research.
Govt mil research = IRINA.
Jehan: Mobile nodes.
Node in region where it’s not supposed to be visible.
All: [Discussion of NV, Jehan work]
Middleware when changing domains.
“Middleware is god”.
Chen: *Heidi to work on this?
Access control issue?
GS: Security makes for a BAD demo.
GIS is good.
Intrusion detection is a good demo.
SM: (1) GIS integration architecture.
(2) Access control – should be a db student.
All: *Whom to do the access ctrl work.
- Work with Jehan.
- Yonghua, Haimin, Ravi, Suba?
SM: Analysis tasks like change detection apply to GIS.
(Talked to Padhraic…)
- Uncertainty model.
Demo – tile display, superimposed maps, demo tool.
NV: *Discuss dissemination another day.
GS: Ad hoc formation of groups.
Have 2 students.
- how to admit to group.
- how to have a secure group (deputizing people).
Gabe: What about u-secure groups.
GS: Charter members.
2nd tier group members.
NV: Adaptive & real time.
Member changes must be resolved.
GS: (1) Don’t change keys.
- may not matter in an emergency if someone has access to old data (via key).
(2) Multi-tier.
- Multiple snapshots.
SM: Access control in XML.
- Hierarchical.
Nitesh?? [Not sure I heard this right – AM]
NV: Connects, disconnects.
CB: In the field, shortage of hardware.
- Free for all, systems down.
- Informal at first
- Then “bring them back” to the computer.
“Build a concrete sand castle on the beach.”
GS: Central groups – locals in middle.
[unclear note – AM] district better.
SM: What if system doesn’t break down.
All: What info is exchanged, what means used.
CB: Design IT part to handle people around w/o machines.
Real world issue.
Example of maintenance people who were barred from a hospital.
Graceful degradation of security.
SM: GLQ – San Diego Gas Lamp Quarter.
SD people…
GS: If just broadcast, no need for a group.
SM: PDAs we’re distributing will have secure spread.
- strong semantic [?] group.
GS: Multicast – heavyweight, not tiny code size.
SM: Throw out XML.
Lightweight XML.
NV: Power aware.
Security, quality, policy tradeoffs.
GS: Can get 1 or 2 students for this.
SM: Use it in SD…
Demo joining, leaving a voice communication.
*Lightweight XML.
GS: Need compelling application scenario.
What is the value added.
Voice – good.
Video – problem. High bandwidth.
All: [Discussion of realistic, needed, practical demo.
Video need argument.
Need for security in exchanging data?
Cooperative scene – role of security?]
CB: Triaging security for critical lines.
Caching; priority – not necessarily security.
Chen: Military vs rescue scenario.
Jehan: Don’t force security when other things fit more naturally.
All: Don’t start with the real practitioners.
NV: 1st responders getting PDAs already in Long Beach.
SM: Video. HP’s vision of rescue future.
Eg, PDAs that automatically translate Spanish voice to English.
CB: “Security is policy by other means”.
BH: Centralized command?
SM: Security – extending a group.
All: [Discussion of 9/11 scenario.]
SM: Continually evolving filter.
All: See starred items above.
Next week mtg on collection and dissemination.
Ram: Talk to Padhraic Smyth.
Yiming, Jehan, + GS Student: ?
Iosif, Xingbo: ? Collection, privacy preserving.
Gloria: ?
Sridevi: ?
AM: Meet with Chen Li re automatically pulling data off Crimemapper.