1. Helsinki University of Technology
Ad hoc networks –
design and performance issues
Juan Francisco Redondo Antón
Master’s Thesis: HUT, Networking Laboratory
Supervisor: Professor Jorma Virtamo
Espoo, May the 28th 2002
Juan Francisco Redondo — Ad hoc networks – design and performance issues

2. Contents Ad hoc networks: features and applications
Capacity: bounds and parameters involved
Medium Access Control
Routing
Simulations
Power control management
Quality of Service
Conclusions / Future work
Juan Francisco Redondo — Ad hoc networks – design and performance issues

3. Helsinki University of Technology
AHNs: features and applications
Why Ad Hoc Networks (AHNs) ?
What are they useful for ?
Fast installation
Dynamic topology
Flexibility
Connectivity
Mobility
Cost
Spectrum reuse possibility
Conferences and meetings
Home environment communications
Emergency search and rescue
Battlefield
Sensors networks with different purposes
(militar, environmental, traffic
sensor networks…)
Juan Francisco Redondo — Ad hoc networks – design and performance issues

4. Capacity of AHNs Theoretical Experimental Bounds on capacity
Paramaters that can modify capacity:
Traffic pattern
Location and mobility of nodes
Range of transmission
Need of common range
Constraints on range: connectivity and throughput
Critical transmission range:
Theoretical values
Possible implementations
Theoretical
Experimental
MAC protocols
Locality of traffic pattern
Effects of relaying
Multipacket reception
Juan Francisco Redondo — Ad hoc networks – design and performance issues

5. Capacity of AHNs Bounds on Capacity of Ad Hoc Networks:
Capacity of wireless networks
2 models of interference
2 hypotheses for the network: random and arbitrary networks
Physical model
Protocol model
Random network
Arbitrary network
Reasons??
Protocol model of interference
An experimental scaling law for ad hoc networks
Juan Francisco Redondo — Ad hoc networks – design and performance issues

6. Capacity of AHNs Parameters affecting capacity in AHNs Traffic Pattern
MAC protocols
Locality of traffic pattern:
[52] uses a power law distribution of the distances to the destinations to show that
- Random traffic pattern is the worst possible
- if  < -2, Cn remains aprox. constant if large enough networks
Effects of relaying: m pure relaying nodes increase capacity like
Multipacket reception (MPR):
- [60] indicates that the use of MPR improves the coefficient of the asymptotic scaling law of AHNs.
- The contribution of MPR is better with high connectivity
- Multipacket transmission (MPT) can also be used
- Some MAC protocols uses MPR: RCT, MQSR…
Juan Francisco Redondo — Ad hoc networks – design and performance issues

7. Parameters affecting capacity in AHNs
Capacity of AHNs
Parameters affecting capacity in AHNs
Location and mobility
Grossglauser and Tse work [58] uses mobility to offer multiuser diversity for the
relaying of packets in AHNs
It is an attempt to facilitate local transmissions
with high probability
Independent movements can attain average
long-term constant source-destination throughput
- Only useful for asynchronous applications
Juan Francisco Redondo — Ad hoc networks – design and performance issues

8. Parameters affecting capacity in AHNs
Capacity of AHNs
Parameters affecting capacity in AHNs
Range of transmission
Need of a common range due to the
necessity of bidirectional links for ACKs
and handshake.
Connectivity and throughput tradeoffs:
- Value of the area and the spatial reuse
- Maximum number of simultaneous transmission-receptions
Juan Francisco Redondo — Ad hoc networks – design and performance issues

9. Parameters affecting capacity in AHNs
Capacity of AHNs
Parameters affecting capacity in AHNs
Range of transmission
Theoretical critical power:
Practical implementations:
COMPOW: a modular solution
An algorithm based on graph
calculation
Juan Francisco Redondo — Ad hoc networks – design and performance issues

10. MAC in AHNs The Medium Access Control Layer in Ad Hoc Networks
Constraints of wireless medium
Transmission technologies: infrared, microwave, spread spectrum
Properties of MAC protocols for AHNs
Proposals:
General for Wireless Networks:
IEEE
HiperLAN
Bluetooth
Specific for Ad Hoc Networks:
CSMA
MACA …
SEEDEX
Juan Francisco Redondo — Ad hoc networks – design and performance issues

11. Routing in AHNs Expected properties: Decentralized execution Loop free
Adaptable to topology changes
Flexible with traffic patterns
Scalable
Bandwidth efficient
Power conservative
Network security
Quality of service support
Metrics:
End-to-end data throughput
Delay
Route acquisition time
Percentage out-of-order delivery
Efficiency
Other metrics
On-demand vs. Table-driven:
AODV / DSR — DSDV
Hybrid schemes as ZRP seems to be the solution for scalability
Protocols designed for high mobility: DREAM, LAR, B-Protocol
Juan Francisco Redondo — Ad hoc networks – design and performance issues

12. Simulations of connectivity
Connectivity and range of transmission
Our goal is estimating the probability of having a fully connected network
as a function of the transmission range
Probability of
fully connected
network
Range of transmission
Juan Francisco Redondo — Ad hoc networks – design and performance issues

13. Power control in AHNs
A power-conservative design affects every network layer:
PHY: Quality of reception
Design of HW in wireless interfaces
Logical Link Control (LLC): accommodating error control schemes to
Application Layer: SW implementation
Traffic requirements
Channel conditions
MAC:
IEEE : allows nodes to sleep temporally through synchronization processes
DBTMA-Enhanced: uses a “busy tones” channel to manage power control
PCMA: extents the handshake procedure to incorporate power-control information
PAMAS: avoids overhearing of the channel to save power
Power-aware routing:
“Energy as a metric” is the crux of the matter
Solutions:
GAF: uses geographic information to make the nodes coordinate themselves
to sleep in turns depending on design parameters
SPAN: creates a backbone of nodes that guarantees routing operation while other nodes sleep
LEAR: looks for balanced energy consumption among nodes
Energy/packet
Cost/packet
Time to network partition
Variance in node power levels
Cost/node
Juan Francisco Redondo — Ad hoc networks – design and performance issues

14. Quality of Service in AHNs
An ad hoc oriented view of QoS includes:
QoS models defines which are the goals:
IntServ / RSVP
DiffServ
FQMM, a QoS model for AHNs
QoS signaling reserves resources:
dRSVP: adaptive adjustment of the QoS level
INSIGNIA: in-band signaling for AHNs
QoS routing finds the QoS routes:
CEDAR
Ticket-based routing
QoS Medium Access Control must
to complete this framework
Support reliable unicast communication
Provide resource reservation
Juan Francisco Redondo — Ad hoc networks – design and performance issues

15. Conclusions / Future work
AHNs are the suitable solution for certain contexts
Capacity is the restraining factor, especially with high number of nodes
Separate analysis of factors impacting capacity provides ideas to increase it
The range of transmission is a critical parameter and has multi-layer influence
MAC problem is open
“Classic” routing protocols are mature
Power control is essential
QoS is an awkward challenge
Integration of network layers
Other questions:
Security
Addressing
Commercially available?
Sharing resources
Juan Francisco Redondo — Ad hoc networks – design and performance issues