VII California Team Meeting

1
VII California Team Meeting

• June 22, 2007
• MTC

2
Agenda Items

• VII California ITSA presence
• UPA and VII California
• Summary of VII California Website
• Demos
• 12 July, Dutch DOT
• 24 July, Three TRB committees, US DOT
• Monitor program and health status of RSE
• Upgrading VII California to National Architecture
  
• Wireless Silicon Valley
• VII tolling application
• DSRC performance Measures
• Installation progress and plans

3
Thanks for VII California ITSA presence

• Photos taken by Bill Stone and Jay Sullivan
• (PATH Pubs)

4
VII California Booth
5
Cabinet Display
6
VIICA Banner and Posters
7
Screen
8
Urban Partnership and VII CA

• Bay Area selected as one of the finalists
• VII tolling has prominent role
• Meetings today in DC to discuss

9
VII California Website

• Art Chitamitara
• www.viicalifornia.org

10
ITSA Premiere

• Whats new?
• VII CA video (homepage)
• Additional new logos
• Dynamic Maps and working RSUs
• Bay Area 511 System (related projects)
• Can be found through Google
• What can we look forward to?
• More VII-related publications, meeting ppt and
  minutes, etc.
• Please visit www.viicalifornia.org

11
(No Transcript)
12
(No Transcript)
13
July Demos
14
July Demos

• July 12th
• Afternoon visit by Director-General
  Rijkswaterstaat (head of Dutch DOT)
• Seeking volunteer VII rides
• July 14th
• Three TRB committees and US DOT demos of  gt 150
  people

15
Demonstration for TRB Workshops

• July 24, 10 am 230 pm
• TRB Committee Workshops at Doubletree Hotel, San
  Jose Airport
• Traffic Signal Systems Comm. (TSSC)
• Intelligent Transportation Systems (ITS) Comm.
• Vehicle-Highway Automation (V-HA) Comm.
• TRB providing shuttle bus service between hotel
  and demo sites on El Camino

16
Workshop Calendar

• Sunday, July 22 TSSC Subcommittee Meetings
• Monday, July 23 Workshop on VII and CICAS,
  mainly presentations (TSSC)
• Tuesday, July 24 VII California Demos, then
  joint workshop by ITS and V-HA Committees starts
  230 with introductory presentations
• Wednesday, July 25 ITS and V-HA workshop
  breakout discussions on research needs in V-I
  Cooperation
• Thursday, July 26 ITS and V-HA workshop
  concludes by 1200 with plenary session to
  summarize breakouts, and presentations from
  Europe and Japan

17
Workshop Attendees

• Registration limited to about 150
• Latent demand for information about VII could be
  strong
• Committee member diversity
• Federal, state and local government
• Consultants and academics
• Industry (more infrastructure than vehicle)
• Domestic and international
• Strong expertise in traffic signal control, ITS
  and vehicle control and safety systems
• Policy and planning interests as well as
  technology

18
Demo Contents

• Real-time 511 data received by portable OBE on
  shuttle bus, displayed via laptop (101 at San
  Antonio, University, Marsh)
• CICAS-V demo by DCX at 5th and El Camino (static
  vehicle, open 2070 cabinet)
• Samtrans bus shuttles visitors along El Camino
  between 5th, Page Mill, showing
• Transit signal priority
• Arterial probe data display (based on a separate
  probe vehicle driving El Camino between bus runs)
• Forward collision warning
• Static display at Page Mill open 170 cabinet
  and signal sniffer, possible to add OEM VII
  vehicles

19
Demo Logistics

• Six groups of 25 visitors, arriving by shuttle
  bus at
• 5th and El Camino at 1000, 1130 and 100
• Page Mill and El Camino at 1040, 1210 and 140
• Each group rides Samtrans bus in one direction
  along El Camino, returning to hotel by shuttle
  from opposite end of demo route, departing from
• 5th Ave at 1130, 100 and 230
• Page Mill at 1050, 1220 and 150

20

• Monitor Program and Health Status of RSE
• Joel Vanderwerf, Ashkan Sharafsaleh
• 6-22-2007

21
Outline

• Backhaul
• Status of RSE
• Status of Page Mill / ECR
• Diagnosis 1
• Diagnosis 2
• Actions to fix the problems
• Workaround

22
Backhaul

• California/ECR (T1 line) has been OK.
• In general, backhaul has been much more reliable
  recently

23
RSE Status

• Page Mill problematic (next slides)?
• Quarry, 101/Univ, Curtner, Charleston
• WRM not responding
• Probably bad fiber converter
• We've seen these symptoms before
• Quarry and 101/Univ are priority for demos
• CPU fan problem at Stanford (minor)?

24
RSE Status Page Mill

• OEM drivers report low signal strength
• Ping reports high packet loss on wire between
  PC-104 and WRM
• Telnet times out, usually
• Not responsive to WRM config requests
• Packet loss on wire is 0 at most sites
• Stanford has similar problem, but less severe
• We believe 1. and 2. are separate

25
Packet loss at Page Mill
Red is result of 100 pings at 1 sec period,
measured every 5 mins Green is average over last
2 hours Note change at time of service, Thursday
morning
26
Note end time
27
Page Mill Diagnosis, 1

• Cables, converters, and WRM seem OK
• We swapped each component
• No anomalies in CPU, mem usage, etc.
• Temperature is a little high

28
Page Mill Diagnosis, 2

• Problem related to two things
• Presence of UDP broadcasts
• phases or signage
• Even at low rate, 2/sec (verified with tcpdump)?
• ping fails only while broadcasts are being sent
• When broadcasts stopped, packet loss is very low
• Time of day
• worst between 8am and 8pm
• but the pattern changed after we worked on it

29
Page Mill Actions

• Check Ethernet card in PC-104
• Half duplex mode?
• Swap in another pc104 (or laptop for testing)?
• Workaround
• Set up sniffer at California/ECR intersection
• The WRM at California has been very reliable
• Also, signal strength is better than Page Mill
• We even had to reduce power
• Goal redundancy for demos

30
VII California Monitor
31
What's new?

• LM Sensors (on pc104 board)?
• More detection of network problems
• More detection of software problems
• Recent signal phase broadcasts
• Graphs using RRDTool
• Web site improvements

32
LM Sensors

• pc104 has on-board sensors
• temperature, fan, voltage
• Linux kernel module reads sensors
• Data is sampled into monitor database
• Sent to monitor server at PATH
• at low data rate
• Aggregated and plotted using RRDTool
• Installed on all 12 field RSU hosts

33
Variables Observed
Software CPU usage broken down by each viica
process, other, and idle memory usage broken down
by each viica process, other, and free config
and status of viica sw internal diagnostics of
monitoring system Hardware (pc104)? voltage fan t
emperature (3 sensors)? clock skew disk
usage uptime

• Network
• backhaul ping
• backhaul ssh
• byte rates (broken down by DSRC/backhaul, in/out
  bound)?
• backhaul capacity (in/out)?
• beacons received (how many andfrom which RSUs)?
• recent signage and other public broadcasts
• recent signal phase broadcasts
• WRM
• wrm config and state
• packet loss between pc104 and WRM

34
Corrective Actions

• Restart signal sniffer
• If phase data not detected, OR
• Phase countdown times are out of bounds
• Restart viica programs
• If crashed or unresponsive
• Send wakeup packets to WRM
• In case WRM decides host is on air interface
• Send channel/power packets to WRM

35
Monitor web site main page
Does host request signage?

• http//path.berkeley.edu/vii/monitor

Backhaul packet round trips
Does WRM respond correctly?
Green, yellow, red depending on age of last report
Can ssh connect?
Links
36
Detail pages
Radio settings
Public message broadcasts
Signal sniffer output
37
Full report page
Detail on ping cmd and results
How many beacons from which source
Total cpu times
For Google map
Monitor diagnostics
38
Graphs using RRDTool

• RRD Round Robin Database
• Widely used by sysadmins
• Stores last N samples of data
• Aggregates (avg, min, max)?
• Derivatives (convert byte counts to rates)?
• Can archive older data with less fidelity
• weekly maxima for last year, plus
• 5 minute averages for last two weeks
• Generates graphs from database

39
An example RRD CGI page
User enters params and presses Submit
40
Autoscaling
Transparency
Stacking
Stacking
Readable times/dates
41
What can we use it for?

• Visualized history of problem at Page Mill
• Discovered similar problem at Stanford
• Discovered CPU fan problem at Stanford
• Diagnosing memory usage problem
• Verify that there's no backhaul hog
• Present ssh/ping history graphs to ISP
• in case we need to complain...
• Anticipate failures

42
Problem automatically detected and process
restarted
Alarming trend starts herecheck logs and version
history!
Alarming trend starts herecheck logs and version
history!
43
To do

• Identification of error states
• Trigger alert emails and RSS feed
• Need to carefully consider...
• How to ignore brief glitches and small errors
• Don't spam!
• More general CGI script
• User can plot any variables for any sites against
  each other, for example
• Correlate temperature with network failures
• Compare backhaul bandwidth at T1 sites

44
Upgrading VII California to National Architecture
45
(No Transcript)
46
SDN Network Access Point

• Checkpoint IPS-1
• Cisco 3845 Router
• Cisco 3845 I/O Module 1-port T3 (for Access/BBN)
• Cisco 3845 I/O modules ( varies based on
  backhaul)
• Cisco ASA 5510 Firewall
• Cisco Catalyst 2960 Switch 24 port
• Cisco Catalyst 3560 Switch 24 port
• Cisco SFP Ethernet Transceiver
• Cisco Smartnet for 2960 24 port
• Cisco Smartnet for 3560 24 port
• Cisco SMARTnet for Router
• Various software license upgrades
• Grand Total 75K

47
SDN Servers

• Dell Poweredge 2950 Server
• NIC - dedicated for IDS
• VMware Server Support
• Windows Server 2003 Enterprise R2 OS
• Red Hat Linux ES4 OS
• Oracle 10g Enterpise R2 DBMS
• Geodetics Commlink Proxy, Version 3.5 (NTRIP)
• Securent Entitlement Management Service
• Sonic MQ Software (Enterpise OS License)
• Grand Total 70K

48
(No Transcript)
49
Wireless Silicon Valley
50
Site Selection Sand Hill or Quarry

• Sand Hill Installation
• Mounted on the large mast arm under a big tree
• The nearest lampposts nearby are also under big
  trees
• Quarry Installation
• Mounted on top of a B-Pole
• Has clear lines of sight to a few lampposts IDd
  by WSV
• Less than 100 meters from this installation

51
A Google photo of Sand Hill and Quarry
Intersections on SR 82
52
Pictures of Quarry Installation
53
Pictures of Sand Hill Installation
54
Our Choice for WSV Quarry Installation

• Clear line of sight
• Ease of retrofit
• A Technocom RSU will replace the existing PATH
  RSU at this location

55
VII Tolling Application
56
VII Tolling Application

• Candidate site for deployment selected
• Dumbarton Bridge initially
• Development will start next week
• Because of 511 emergency operation priority,
  schedule will slip
• Scheduled to be complete in October
• Coordination with BATA and ACS

57
National VII Tolling - Update

• Effort underway by OmniAir, sponsored by USDOT
• Develop a Concept of Operations
• Supports the POC in SE Michigan
• Web Conference June 29th
• Technical Memo due out today

58
National VII Tolling Tech Memo

• Includes Storyboards
• Current tolling functions and requirements
• Proposed VII tolling concepts
• Basic functions
• Configuring an existing toll zone or toll lane
• Activating a vehicle for VII tolling
• Opening a VII account
• Paying a toll in a multi-protocol environment.

59
Field Measurements of Vehicle to Roadside
Communication Performance

• Susan R. Dickey

60
Equipment Set-Up

• Assembly of three gumstix netduo expansion boards

• Antenna placement with 8 active OBUs on one
  vehicle

61
Location of RSU

• Test 1
• Fixed loads at receiver
• RSU _at_ 6th and Brannan in San Francisco
• Test 2
• Fixed sending interval
• RSU _at_ 3rd and Townsend in San Francisco

62
Experiment Results Test 1 (1)

• Average of remote and local RSSI by GPS location
• The recv RSSI is higher when vehicle is closer to
  RSU

63
Experiment Results Test 1 (2)

• Probability of successful round trip transmission
• as a function of the interval between packet
  sends, for 1, 2, 4 and 8 nodes
• with varying packet sizes of 200, 400 and 800
  Bytes

64
Experiment Results Test 1 (3)

• Time (in seconds) until the first 8000 bytes has
  been transmitted
• as a function of the interval between packet
  sends, for 1, 2, 4 and nodes
• with varying packet sizes of 200, 400 and 800
  Bytes

65
Experiment Results Test 2 (1)

• Transaction delay for 8000 byte transaction as a
  function of data rate
• for 9 nodes, packet size 200 bytes
• with various testing routes, 10 millisecond
  interval between packet sends
• Transaction jitter (standard deviation of delay)
  as a function of data rate
• for 9 nodes, packet size 200 bytes
• with various testing routes, 10 millisecond
  interval between packet sends

66
Experiment Results Test 2 (2)

• Transaction delay for 8000 byte transaction as a
  function of data rate
• for 9 nodes, packet size 1400 bytes
• with various testing routes, 10 millisecond
  interval between packet sends
• Transaction jitter (standard deviation of delay)
  as a function of data rate
• for 9 nodes, packet size 1400 bytes
• with various testing routes, 10 millisecond
  interval between packet sends

67
Analysis (1)
R1 Trans. time decreases as sending interval
increases. Reason If sending interval is too
short (rate is too high), there are many
collisions. R2 Trans. time increases as sending
interval increases. Reason Sending interval is
already long enough for contention resolution.
Thus, channel is not well utilized (bandwidth
lost). There is an optimal sending interval.
68
Analysis (2)
R1 Prob. of successful TX increases (up to a
bounded prob. p) as sending interval
increases. Reason As sending interval increases,
there are fewer collisions. R2 Prob. of
successful TX approaches a limit when sending
interval keeps increasing. Reason Sending
interval is already long enough for contention
resolution. Thus, prob. of successful TX will not
increase instead, it is bounded by p, which is
decided by PHY layer performance.
69
Future Work

• The disadvantages of application layer only
  studies are obvious
• Physical layer effects that may be contributing
  to poor performance must be inferred indirectly.
• It is necessary to be able to integrate data
  acquired at all layers of the network stack for
  complete understanding.
• We plan to use this equipment to carry out
  further vehicle to roadside and vehicle to
  vehicle tests at VII California RSUs, in
  conjunction with more physical layer testing.

70
Installation Progress and Plans

• Ashkan Sharafsaleh

71
Design and Installation

• Surveying of the potential sites is in progress
• Dumbarton bridge has been selected for the toll
  application
• More sniffers will be built
• Further research is needed for intersection sites
  with 170 controller
• Need to built a computer that relays the sniffer
  data to MCNU Technocom RSU on the signal pole

72
Procurement Process

• PATH is working to procure the following for
  the future installations
• MCNU Technocom RSUs
• Fiber and power cables
• NEMA boxes, antennas, brackets, different cables,
  different connectors, and so on
• Fiber converters and GPS units