Transcript Document

Scuola Superiore Sant’Anna
A possible roadmap to enter
the WSN research community
Paolo Pagano
ReTiS projects for WSNs (1/2)
•
Real-Time operation of Wireless
Sensor Boards:
– Using the Crossbow MICAZ sensor
board, the most popular in the
research domain;
– Erika E. ported to the AVR platform
(the microcontroller mounted on the
MICAZ board);
– Implementation of Real-Time
protocols for time synchronization
and localization.
•
Deployment of WSNs for:
– Airflow monitoring in agriculture;
– Critical bio-medical applications.
•
Simulation:
– NS-2/RTSim co-simulation to
validate communication protocols in
diversified scenarios.
ReTiS projects for WSNs (2/3)
NicKName
Participation
Funding
Issues
Last
Discussed
INTERNAL
SSSA labs
No
Data managing in WSNs;
Granularity of information;
Data and computing model;
Intelligent Water Management.
Pontedera
13/12/06
AGRO
F.Mencarelli
(UniVT)
P.Bruschi
(UniPI)
No
tinyOS-based nonRT operations;
Model of Communication;
Time synchronization (?);
Energy management (duty cycles, etc.);
Sensor devices (market products or
research prototypes?)
Pisa
18/12/06
ART_DECO
FIRB
Yes
Middleware not better specified.
Bolzano
19/12/06
RIMACS
EU
Yes
Agent-based mechatronics control
Pisa
technology with embedded intelligence;
October??
Distributed (Wireless) Networked systems
17/07/2015
Paolo Pagano
ReTiS
people
ReTiS projects for WSNs (3/3)
NicKName
Participation
Funding
Issues
Last
Discussed
WPDRTS
ReTiS
No
OS simulation in NS-2;
QoS in WSN;
Transaction simulation in RTSim;
Data-centric MW in NS-2;
RT constraints of communication.
Pisa
17/11/06
PRIN
SSSA,
PoliTO,
UniUD,
UniPI
Yes
Medical and biological applications for
WSNs;
Asynchronous Digital Device Systems;
QoS in WSN.
Torino
3/11/06
17/07/2015
Paolo Pagano
ReTiS
people
Magnifying lens on MWs
• In several projects we face the problem of
communication in the WSNs;
• In some cases we go further and mention
Distributed Systems, Data Handling, and
Information Management;
• We may answer these questions:
–
–
–
–
17/07/2015
Which MWs exist on WSNs?
What are these MWs designed for?
What can the lab produce in this field?
What’s the status of the art in the selected field?
Paolo Pagano
MWs for WSNs
• From ETH Summer School on WSN:
–
http://www.vs.inf.ethz.ch/events/dag2005/program/lectures/marron-2.pdf
• I learnt there exist 4 categories of MWs:
– Classic middleware:
• As “Impala” focuses on communication primitives, to tune long-running
applications or to update parts of the code (versioning support);
– Data-centric middleware:
• As “TinyDB” abstracts the WSN as a distributed DB: code is immutable, the
net is in charge of periodically sensing the environment;
– Virtual Machines:
• As “Maté” provides ports where pieces of code aka capsules (limited in
number) may migrate through; continuous update of running WSNs;
overhead?
– Adaptive middleware:
• As “TinyCubus”, a framework to wire together services, modules provided
by the OS and by the user to minimize some cost metrics.
17/07/2015
Paolo Pagano
Data-centric RT MW: DSWare (1/2)
• Stankovic group is very active in this field, tens
of publications:
–
http://www.cs.virginia.edu/~stankovic/sensornet.html
• They have a product “DSWare”:
–
http://www.cs.virginia.edu/~stankovic/psfiles/li.pdf
– It’s still at the level of simulation (using GloMoSim) or
even abandoned (the pub is dated 2004);
– 2 relevant features embedded in the service:
• Global scheduler based on distributed EDF (following the
time window of validity for a probed variable) (?);
• Event Detection Service, assigning a confidence level to
each event and plans communication therefore.
17/07/2015
Paolo Pagano
Data-centric RT MW: DSWare (2/2)
•
Some drawbacks:
–
–
–
–
•
MAC layer not defined. Does 802.15.4 exist in GloMoSim?
The MW can schedule transmissions: what assumptions are made upon RT behavior
of layers underneath?
Which assumptions done on the OS interrupt service?
How is the schedule shared?
Quoting the results:
T  T (detection)  T (event)
• Only quoting the
average time and RMS,
not the maximum;
• Range of events is tiny
[0-25].
17/07/2015
Paolo Pagano
Follow-up’s and implementations?
• Follow-up of Event Detection Service is GEM:
–
http://www.cs.virginia.edu/~stankovic/psfiles/binjia-inss.pdf
– A lot of theory, semantics, language definition, Petri Nets;
• Anyhow a military application (VigilNet) such as Target
Tracking is published:
–
http://www.cs.virginia.edu/~stankovic/psfiles/RTAS06-He-1.pdf
– implemented on top of TinyOS but programming directly against
the hardware;
– a lot of blah-blah to say that 80% of processing is done within
the deadline:
17/07/2015
Paolo Pagano
An attempt of summarizing...
• A data-centric RT-MW exists (in simulation and
in dedicated applications);
– The choice of simulator indicates that Stankovic
group ignores all layers up-to the network; the used
MAC may not be the correct one. Looking at the
GloMoSim WEB site:
• “For the MAC layer we have protocols for CSMA, MACA,
and IEEE802.11”.
– Deployments rely on TinyOS and are done
programming against the hardware. Real data
analysis shows puzzling results (not fulfilling at least
hard RT constraints).
17/07/2015
Paolo Pagano
TinyOS interlude (to deprecate TinyOS)
• About the MAC layer, please visit the FAQ:
– http://www.tinyos.net/faq.html#SEC-76
• As far as I understand:
– Micaz mounts CC2420 radio;
– The radio transmits 802.15.4 formatted frames..
– …but the MAC protocol is B-MAC:
• http://www.cs.berkeley.edu/~polastre/papers/sensys04-bmac.pdf
• 802.15.4 standard not exploited, translation
done at the interface between PHY and MAC
layer in order to be transparent in a WPAN.
17/07/2015
Paolo Pagano
What can we do in such a field? (1/4)
• Simulation:
– provide a complete 802.15.4-based simulation of a distributed
system, specifically a layer for RT data-services;
What do we miss?
– Correct processing of IRs and routing messages;
– RTSim support for transactions;
– Protocols for sharing global schedule (Tullio knows the HOWTO
as always );
– Time synchronization protocol;
– An advanced study of 802.15.4 libraries (namely clustering,
routing, etc).
17/07/2015
Paolo Pagano
What can we do in such a field? (2/4)
• Programming the Real Hardware:
– provide a complete 802.15.4-based RT distributed system,
specifically a MW for data service. Even we can do better the
localization case study and apply for RTAS08 .
What do we miss?
Application
• Completion of Network layer
in EE (write and read 802.15.4
datagrams);
• The MicaZ development kit
Antonio ordered 3 months ago.
17/07/2015
Network Stack
Kernel Layer
Hardware Abstraction Layer
CPU
MCU
Board
Layer
Layer
Layer
Paolo Pagano
What can we do in such a field? (3/4)
• An update on Real Hardware
has come meanwhile: Flex
exists now and a GPL version
of ERIKA runs on the
microcontroller;
• do we have a WPANcompliant radio? apart from
reading the sensors, can we
transmit 802.15.4 frames?
• is it worth to write down the
Network stack for dsPIC?
• in other words, can we replace
the XBow by Evidence and
build the “Motes” at home?
17/07/2015
Paolo Pagano
What can we do in such a field? (4/4)
• Going back to the MW:
– we can profit of the expertise developed in similar projects (e.g.
CAMBADA, by L.Almeida et al):
• http://robot.unipv.it/toolleeo/publications/papers/2004/iscis_2004.pdf
• OK, the net is 802.11;
• OK, the robots run LINUX ;
• … but there is an implementation of a RTDB.
– Wireless remote programming eases life!
– … by the way some basics of ordinary high-end RT-DBMS is
uploaded to ReTiS PLONE.
17/07/2015
Paolo Pagano
TinyOS-based project (AGRO/ INTERNAL)
• Purpose of the project:
– Monitoring of temperature (ventilation later on) indoor;
– Synchronizing the motes;
– Exploiting naïve protocols of communication;
– Alarm detection (client/server, distributed system
later on).
• Setting up the test-bed as soon as possible in
order to trigger funding and joint-researches.
17/07/2015
Paolo Pagano
Other issues related to communication
• Usually the energy consumption is quoted with respect
to Communication Cost (Number of messages present
in the network after an event);
– G. Anastasi mentioned that a group Milan is involved in the
study of differential consumption (per layer) in motes; if we dig
in the publications we probably find something interesting.
• No study of transmission. I found on the net a
HandBook about Wireless Telecommunication to study
(uploaded to ReTiS Plone):
– in-door/out-door effects;
– modulation;
– codification.
• A real effective study about 802.15.4 PHY layer is
probably missing…
17/07/2015
Paolo Pagano
Now let’s go to the to-do list…
MANETS?
• Sensor reading in TinyOS;
• Client/Server communication w/
TinyOS;
Tullio• RBS(*) implementation in TinyOS;
Nino • Transaction in RTSim;
Gianluca
• RBS implementation in NS-2;
Tommaso
• Delay correction for IRs in Ns-2;
• Gateways and routing paths in Ns-2
HIGH
(remainders);
• Sensor reading in ERIKA;
• Network primitives in ERIKA;
• Energy management for the motes;
Luca
• NesC-like programming language
for ERIKA.
HIGH
(*) Basics of Time Synchronization are in
ReTiS PLONE
MEDIUM
LOW
Fabio
Anna
Enrico
Antonio
Peppe
Paolo
Mangesh
High level hints?
HIGH
HIGH
• CAP or CFP?
• Emulators for WSNs
• Future meetings must address very
specific and point-like issues instead
of large survey as this one 
• …………….
• …………….
• …………….
• …………….
• …………….
• …………….
MEDIUM
LOW
Peppe
Giorgio
Network Scenarios (1)
• Starting from simple
scenarios:
– 4 nodes located at the
edges of a square
communicating to a sink
placed at the center of
mass; no hidden terminal;
no network structure
(single cluster); no routing;
the sink plays the role of
WPAN coordinator.
• To re-obtain the
governing laws of RTcomputing.
17/07/2015
Screenshot of the Network
AniMator (NAM)
Paolo Pagano
Network Scenarios (2)
• Complicating the picture:
– Introducing routing paths
and data streams;
– Nodes connecting different
clusters fetch and forward
the readings coming from
other clusters.
• How do RT-issues
influence the reactivity
of such nodes?
Screenshot of the Network
AniMator (NAM)
17/07/2015
Paolo Pagano
Conclusions and Outlook
• The WSN projects where ReTiS is involved in have been
presented;
• Most of the attention relies on TinyOS-based applications, NS2/RTSim cosimulator, and on the missing ERIKA-compatible
Network stack for MicaZ board;
• Long term proposals to write a RT event detection service in WSN
running ERIKA;
• An attempt to write down a TODO list;
• A lot of material uploaded to PLONE.
• Visit:
–
http://feanor.sssup.it:8080/retis/ReservedArea/sens_nets/useful-material
• Next meetings:
– on 25th (Philips);
– about Time Synchronization on Feb 1st 3pm.
17/07/2015
Paolo Pagano