Transcript IIT-CNR presentation

First Year Results and Lessons,
and Project Directions for the
Second Year
Marco Conti
Computer Networks Dept., IIT CNR
[email protected]
http://cnd.iit.cnr.it/mobileMAN
Methodology
2200 papers (citeseer), but no mainstream products
More than 10 years of research but this area is far
from being consolidated
Three Areas where Ad Hoc lacks “Realism” (*):
1. Integration (lack of)
2. Implementations/Testbeds (lack of)
3. Experimentation (lack of)
(*)
Christian Tschudin, Henrik Lundgren, Erik Nordström, “Embedding MANETs in the Real
World“, Personal Wireless Communications”, Proceedings of the 8th IFIP-TC6
Conference, Venice, Italy, 23-25 September 2003, Lecture Notes in Computer Science
LNCS 2775.
MobileMAN IAB meeting (Helsinki, 7 Jun 2004)
State of the Art: open issues
1.
Integration (lack of)
Focus was mostly on revising routing Internet Draft. Current ID revision
numbers: v13 (AODV, now an RFC), v9 (DSR), v11 (OLSR), v10
(TBRPF)
Main other areas have been analyzed (and new areas are continuously
added), see (*):
- Most works focus on a single layer/issue (MAC, Routing, TCP,
middleware, etc.); or
- Studies of the interation among few layers (TCP and MAC, TCP
and routing, energy and topology control, …)
Lack of an overall view of MANET architecture and protocols
(*) I. Chlamtac, M.Conti, J. Liu, “Mobile Ad hoc Networking: Imperatives and Challenges”, Ad Hoc
Networks Journal, Vol.1 N.1, 2003.
MobileMAN IAB meeting (Helsinki, 7 Jun 2004)
State of the Art: open issues
2.
Implementations/Testbeds (lack of)
Only limited hands-on experience available
(typically 4 to 10 nodes, up to 37 at Uppsala)
3.
Experimentation (lack of)
Internet Drafts production cycle mainly driven by simulation work:
“implementations can come later, simulations tell the truth”
Few attempts to validate/calibrate simulative studies with measurements
Can we trust the protocols inside the simulators?
MobileMAN IAB meeting (Helsinki, 7 Jun 2004)
Can we trust the protocols inside the
simulators?
Scenario
An IEEE 802.11 network with 50 nodes moving
according to the Random Waypoint mobility model
in a 1500m x 300m closed rectangular area. The
nodes maximum speed is equal to 1 m/sec.
In the considered scenario there are 3 FTP sessions
(1460 bytes per packet), and ten CBR sessions
(512 bytes per packet).
Results
Figures show the sum of the 3 FTPsession throughput by varying the length of
the pause in the Random Waypoint model
Results are highly dependent on the
simulation tool.
MobileMAN IAB meeting (Helsinki, 7 Jun 2004)
Can we trust simulative results?
2. No calibration of simulators with
real measurements
3. Not all 802.11 aspects are
modelled, e.g., most simulators
assume the same Tx range for
unicast and broadcast frames.
Brodcast Range
Unicast Range
GW
real
FS1
FS2
Gray Zone
MS
The Communication Gray Zones
Problem
MobileMAN IAB meeting (Helsinki, 7 Jun 2004)
MobileMAN: Experimentation &
Implementation
Experimentation
Use of real testbeds:
•
to validate the developed solutions whenever possible or, (at least),
•
to calibrate/validate the simulation models to be used for
architecture and protocols desgin
Implementations
• Whenever possible, hardware/software implementation of the MobileMAN
solutions
• The integration of the developed solutions in a testbed.
MobileMAN IAB meeting (Helsinki, 7 Jun 2004)
MobileMAN: Integration
The development, and validation, of effective integrated solutions
for the relevant technical issues of self-organizing networks:
routing and medium access control protocols, power management,
security, and location.
application 2
application k
middleware
transport and
network layer protocols
wireless ad-hoc technologies
MobileMAN IAB meeting (Helsinki, 7 Jun 2004)
cooperation
model
power management
application 1
MobileMAN Status
Experimentation
•
We will see some experimental results in the following
presentations
Implementation
•
The main activity of the second year of the project
Integration
•
A preliminary integration will be performed in the last months of the
II year.
MobileMAN IAB meeting (Helsinki, 7 Jun 2004)
MANET Architecture
MANET IETF WG proposes a view of mobile ad hoc networks as an
evolution of the Internet [1]:
• IP centric view (“solve mobile routing problems at the IP layer”), and
• use of a layered architecture
Advantages: It guarantees that the TCP/IP protocol stack can efficiently
operate on a large set of heterogeneous technologies.
Drawbacks: layers independence makes difficult to fix some ad hoc
networks, see for example TCP-MAC interaction problems.
MobileMAN IAB meeting (Helsinki, 7 Jun 2004)
MANET reference architecture
Application 1
Application 2
Application n
Inter-layer
tasks
Middleware
Applications
&
Middleware
Service location, Group communication, DSM
Security
Transport and Network layer protocols
Networking
Cooperation
TCP, IP routing, Forwarding, Addressing, Interconnection
Energy
Saving
Enabling
Technologies
802.11
Bluetooth
Hyper LAN
Medium Access Control, Antennas, Power Control
MobileMAN IAB meeting (Helsinki, 7 Jun 2004)
Cross layers’ Dependencies in MANET
Design
Efficient solutions for ad hoc networks require a more strict cooperation among
protocols belonging to different layers [2]:
for example, from the energy management standpoint, power control and multiple
antennas at the link layer are coupled with scheduling at MAC layer, and with energyconstrained and delay-constrained routing at network layer
The question is to what extent the pure layered approach needs to be
modified?
• Extension of the layered architecture: layer triggers
Layer triggers are pre-defined signals to notify some events to the higher
layer, e.g., failure in data delivery.
• “Violation” of the layered architecture: full cross layering design
Neglecting the layers separation principle by a joint design of the protocols
[2] A.J. Goldsmith, S.B. Wicker, “Design Challenges for Energy-Constrained Ad Hoc Wireless Networks”, IEEE
Wireless Communications, Volume 9, Number 4, August 2002. pp. 8- 27.
MobileMAN IAB meeting (Helsinki, 7 Jun 2004)
Cross Layering Pros
Cross Layer optimization for all network functions. Cross layering is a
must for functions such as energy management, but provides benefits
for all network functions.
Both local and global adaptation can be performed to adapt the system
to highly variable ad hoc network conditions, and to better control the
system performance
Full Context Awareness at all layers. At each layer, protocols can be
designed to be aware of the network status, energy level, etc. Cross
layering makes easy to achieve context awareness at the application
layer.
Reduced overhead for collecting the network status information avoiding
data duplication at different layers.
MobileMAN IAB meeting (Helsinki, 7 Jun 2004)
Cross Layering Cons
• Protocols re-design (?)
 minimal re-design is desirable
• No Layers independence (?)
 “spaghetti” style in protocols design must be
avoided
Vikas Kawadia, P.R. Kumar, "A Cautionary Perspective on Cross Layer Design",
Technical reports
MobileMAN IAB meeting (Helsinki, 7 Jun 2004)
MobileMAN Cross Layer Architecture
Applications
Network status
Cooperation
Performability
Power Management
Middleware
Socket API
(Simplified) Transport Protocol
Forwardin
g
Users’
location
Network Layer
Enhanced Wi-Fi cards
Routing
the kernel of the
architecture is a shared
memory (“Network
Status” in the figure)
that is a repository of all
the network status
information collected by
the network protocols.
All protocols can
access this memory to
write the collected
information, and to read
information
produced/collected
from the other
protocols.
MobileMAN approach to cross layering tries to balance performance
optimization with protocols’ stack adatability
MobileMAN IAB meeting (Helsinki, 7 Jun 2004)
MobileMAN Cross Layering
Architecture
Allows for a full compatibility with standards, as it does not touch the core
functions of each layer.
Is robust to upgrading, and protocols belonging to different layers can be
added/removed from the protocol stack without modifying the operations at
the other layers.
For example, using the legacy TCP protocol as the transport protocol of the MobileMAN
architecture only imply that cross-layer optimizations will not occur at this layer.
It maintains all the advantages of a modular architecture.
Layer separation is achieved by standardizing the access to the Network Status.
MobileMAN IAB meeting (Helsinki, 7 Jun 2004)
Directions for the Second Year
1. Continue along the construction of a MANET
following the IETF approach to be used for
experimental tests (internal and with users)
2. Refine the MobileMAN Architecture and Protocols to
include the Cross Layering principle
MobileMAN IAB meeting (Helsinki, 7 Jun 2004)
MobileMAN protocol stack
architecture with Cross Layering
Cross-Layering
NeSt
Ring overlay for
subject-based
routing
Socket-based
networking
NeSt API
Common API
Middleware
(Subject-Based Routing)
Berkley Sockets (stream, datagram)
Transport
(TCP, UDP)
Network
(Link-State routing, IP forwarding)
IEEE 802.11 (MAC + Physical)
MobileMAN IAB meeting (Helsinki, 7 Jun 2004)
MobileMAN Cross Layering
Architecture: II Year Open Issues
•
How should the network status information managed?
•
How protocols belonging to different layer access to the common
information?
•
Make protocols cross-layer aware
•
Protocols overheads cannot be evaluated in isolation but new crosslayer metrics need to be defined.
MobileMAN IAB meeting (Helsinki, 7 Jun 2004)
NeSt Hypothesis
We have a language, that the NeSt and the protocols can
use for:
Describing data
Writing functions to analyse/monitor data
The NeSt manages abstractions of data and events,
internally maintained by protocols
Example: topology information collected at the routing layer is
exported in a graph structure in the NeSt
MobileMAN IAB meeting (Helsinki, 7 Jun 2004)
NeSt functionalities
NeSt groups and controls all cross-layer interactions
It implements two models:
Synchronous interactions, to allow data sharing among protocols
Asynchronous interactions, to allow network event subscription and
notification
MobileMAN IAB meeting (Helsinki, 7 Jun 2004)
Registering and seizing of an
abstraction
•register : (void)  ProID
•seize : (ProID, AbsID, CallBack())  result
NeSt
register()
a. Register
seize()
1. Generate PID
2. Return PID
P1
access()
subscribe()
1. Verify callback
2. Grant or reject
seizing
MobileMAN IAB meeting (Helsinki, 7 Jun 2004)
notify()
monitor()
b. Seize an abstraction
with PID, AID
and callback
P2
Synchronous interaction
•access : (ProID, AbsID, Filter())  result
NeSt
register()
a. Access an abstraction
with PID, AID and Filter()
seize()
P1
access()
1. Invoke callback
for AID
subscribe()
2. Apply Filter()
on AID
notify()
3. Return Filter()
result
monitor()
MobileMAN IAB meeting (Helsinki, 7 Jun 2004)
INTERNAL DATA
P2
b. Abstract internal
data in NeSt format
Asynchronous interaction:
internal events
•subscribe : (ProID, EvID)  result
•notify : (ProID, EvID, EvInfo)  result
NeSt
1. Register
subscription
(PID, EID)
register()
seize()
P1
access()
1. Check EID on the
subscription list
subscribe()
2. Deliver the
occurrence to
matching
subscribers
notify()
MobileMAN IAB meeting (Helsinki, 7 Jun 2004)
monitor()
a. Subscribe event with
PID and EID
Catch e
P2
b. Notify event with
PID, EID and info
regarding the
occurrence
Asynchronous interaction
•monitor : (ProID, AbsID, Monitor())  result
NeSt
register()
a. Monitor event with
PID, AID and Monitor()
seize()
1. Verify Monitor()
2. While (true)
2a. Invoke callback
for AID
2b. Apply Monitor()
to AID
2c. If match found
notify PID
MobileMAN IAB meeting (Helsinki, 7 Jun 2004)
P1
access()
subscribe()
notify()
monitor()
Internal data
P2
b. Abstract internal
data in NeSt format
Second Year Objectives
Wireless Technologies
Include the cross-layering view at datalink layer.
Design and implement a full datalink layer that includes the enhanced MAC
protocol for ad hoc networks as designed during the first year.
Networking services
Include the cross-layering view at network/transport layer
Design/Implementation of packet forwarding schemes suitable for MobileMAN.
Development and testing of an ad hoc network
Co-operation Model
Implementation and validation of the cooperation models and mechanisms
defined during the first year.
MobileMAN IAB meeting (Helsinki, 7 Jun 2004)
Second Year Objectives (cont.)
Middleware
Adaptation of Pastry middleware for the MobileMAN environment to exploit cross
layering.
Development of new solutions for P2P information delivery based on Pastry.
New applications and services
Include the cross-layering view at application layer
Adaptation of existing applications, identified, during the first year, to run on top of
MobileMAN.
Socio-economic Model
Apply the developed methodology for evaluating social, anthropological potential of
MobileMAN, and provide the results for improving the technical parts.
To investigate market-based mechanisms for exploiting the self-organised paradigm
MobileMAN IAB meeting (Helsinki, 7 Jun 2004)
Questions ?
References
MobileMAN: Deliverable D5
M. Conti, S. Giordano, G. Maselli, G. Turi, “Cross Layering in MANET: the
MobileMAN Approach”, IEEE Computer, February 2004.
M. Conti, J. Crowcroft, G. Maselli G. Turi, “A Modular Cross Layer
Architecture for Ad Hoc Networks”, in “Theoretical and Algorithmic Aspects
of Sensor, Ad Hoc Wireless and Peer-to-Peer Networks”, Jie Wu (Editor)
CRC Press, 2004
Thank You !
MobileMAN IAB meeting (Helsinki, 7 Jun 2004)