presentation
TRANSCRIPT
Real-time monitoring and delay management of a
transport information system
Lorenzo Sfarra
Master Degree Thesis 2013/14
Dott. Ivano Malavolta Dott. Mattia D’Emidio
Urban/Rural population2009
49% 51%
Urban Rural
2050
30%
70%
Urban Rural
+84%
Source: UN, “Urban and Rural Areas 2009”, “Megacity Challenges 2006”
• Traffic Congestion: 1-3% less of GDP (Gross Domestic Product)
• ➥ Efficient public transport: how?
Intelligent Transport System
!
• Application of a combination of technologies to the existing transport system!
• provides safer, quicker travel with enhanced mobility
• Example: • Variable speed limits
BusOnAir• Focus: public bus transport in L’Aquila • DISIM project since 2010, on data provided by AMA
Bus Stop
On Board
Web
Mobile
Services
Goals
• BusOnAir1 to BusOnAir2!
• From SATE to Fully Time Expanded
• Multi-days support (7)
• Real-time, delay management
• Performance improvements
Timetable
• Set of buses Z, set of stations B
• Set of elementary connections C, whose elements c are 5-tuples of the form c = (z, S1,S2, td,ta):
• bus z leaves station S1 at time td;
• the immediately next stop of bus z is station S2 at time ta.
S1 S2td ta
z
Important ProblemsEAP
AB
• Earliest Arrival Problem • Input: (A, B, t0) • Output: earliest possible
arrival time t2 at B, departing from A at time t1 ≥ t0, minimising the difference between t2 and t0
MNTP
• Minimum Number of Transfers Problem
• Input: (A, B, t0) • Output: arrival at B,
departing from A minimising the number of bus transfers…….
BICRITERIA (X,Y)
Models (1/2)
• BOA1: Simplified Acyclic TE!
• 1 day, acyclic
• td and ta ∈ [0, 1439]
• shortest-path on TE ➡ optimal EAP
StopStation
Route
Run • Time Expanded (TE)*!
• + CheckPoint!
• stay-edges and train-edges
* Ford & Fulkerson “Constructing maximal dynamic flows from static flows”, 1958
Models (2/2)• ➡ BOA2: Fully Time Expanded!
• From 1 to N supported days (design choice: 7)
• Overnight edges between the i-th day and the (i++1)-th day: at day i, t = td + i ⋅1440
• The N-th day does not have overnight edges (DAG)
Day 1 … Day 7
Delay Management3
⧸3
53
⧸⧸
• Users can miss a bus transfer or a bus can be late!
• New bus transfers are possible: buses A,B,C in temporal order at station S. If a delay occurs for A, the order can change in B,A,C or B,C,A
!• Stop: staticTime and time attributes
3
Delay Management3
⧸
53
⧸⧸
• In case of a delay D in the bus stop S: 1. update time(S) = time(S) + D
• time(S) could be the same as staticTime(S) 2. Reflect the delay updating the graph (Run) 3. update the stop position ⟹ has always to be
consistent according to the temporal order of all the stops in a station • setNextInStation(n), setPrevInStation(n)
Real-time trackingUDP!
Server
lat: X long: Y IMEI: I signal: S speed: V dd/mm/yyyy HH:MM
BusOnAir Server
/runs/{id}/rt/updateposition?time=xx&lat=X&long=Y
Performance Improvement
File Buffer !caches data as stored on the durable media, improves read/write performance by writing to the cache and deferring durable write
Objects cache!!
caches nodes and relationships in a form optimised for fast traversal of the graph
5-10x faster than File Buffer
Cache BoA API: /init/
⤵⤵
• write operations: 0.5~1 second faster
0
1,75
3,5
5,25
7
Res. 1 Res. 2 Res. 3
BusOnAir 1 BusOnAir 2
BoA1 BoA2 AVG
Res. 1 5.128 3.469 0.124
Res. 2 6.164 3.286 0.180
Res. 3 4.681 3.269 0.173
Why still seconds? TransientStops
Filled cached1st query
Performance Improvement
Conclusion• BusOnAir1 to BusOnAir2!
• From SATE to Fully Time Expanded
• Multi-days support (7)
• Real-time, delay management
• Performance improvements
• Future Work!
• Improve performance with new more effective models
• graph models with speed up techniques (Dynamic Timetable Model) or array-based (CSA)
• APIs security
• new website interface, mobile app
Thank youReal-time monitoring and delay management of a
transport information system
Lorenzo Sfarra <[email protected]>