Transcript CECS470 - WINSLab

Wireless Mesh Networks
Myungchul Kim
[email protected]
A Survey on WMNs
- IEEE Communications Magazine, pp. 523-530, Sept 2005
• Introduction
–
–
–
–
–
Mesh routers and mesh clients
Multi-hop communications with much lower transmission power
Mesh router with multiple wireless interfaces
Mesh client with a single wireless interface
Customers without wireless NICs can access WMN through for
example, Ethernet
– Advantages: low up-front cost, easy network maintenance,
robustness, reliable service coverage, etc.
– The available MAC and routing protocols are not scalable;
throughput drops significantly as the number of nodes or hops in
WMNs increases.
Network architecture and critical design factors
• Network architecture
- Infrastructure/backbone WMNs
- Fig 1.
Network architecture and critical design factors
- Client WMNs
- Actually the same as a conventional ad hoc network
- Routing and self-configuration
- Hybrid WMNs
- Fig 2
Network architecture and critical design factors
- The characteristics of WMN
- Supports ad hoc networking: self-forming, self-healing and selforganization
- Multi-hop wireless networks
- Minimal mobility and dedicated routing and configuration
- Mobility of end nodes
- Mesh routers integrate heterogeneous networks, wireless and
wired
- Power-consumption constraints are different
- Compatible and interoperable with other wireless networks
Network architecture and critical design factors
• Critical design factors
- Radio techniques
- Directional and smart antennas, multiple input and multiple output
(MIMO) systems, and multi-radio/multi-channel systems
- Refconfigurable radios, frequency agile/cognitive radios, software
radios
- Scalability
- Mesh connectivity: network self-organization and topology control
algorithms are needed
- Broadband and QoS
- Security: no centralized trusted authority to distribute a public key in
a WMN
- Ease of use
- Compatibility and interoperability
Advances and research challenges
• Network capacity
- The guideline to improve the capacity of ad hoc networks: a node
should only communicate with nearby nodes [1].
- Througput capacity can be increased by deploying realying nodes. ->
utilizing the node mobility -> transmission delay and buffer
- Nodes need to be grouped into clusters.
• Layered communication protocol
– Physical layer
• Advanced physical-layer technques: congnitive radios dynamicaly
capturing the unoccupied spectrum
• Enable the programmability of all advanced physical layer techniques
• Open research issues: improve the trasmission rate and the performance
of physical-layer techniques and utilize the cross layer design between
MAC and the physical layer
Advances and research challenges
– MAC layer in WMN
•
•
•
•
•
More than one-hop communcations
Distributed for multipoint-to-multipoint communication
Network self-organization
Mobility is low
Single-channel MAC and Multi-channel MAC
– Single-channel MAC
• Modifying existing MAC protocols
• Cross-layer design: directional antenna-based MACs and MACs with
power control -> hidden terminal problem
• Proposing innovative MAC protocols: how to design a distributed
TDMA MAC protocols overlaying CSMA.CA?
Advances and research challenges
– Multi-channel MAC
•
•
•
•
Multi-channel single-transceiver MAC
Multi-channel multi-transceiver MAC
Multi-radio MAC
Open research issues: scalable MAC, MAC/physical cross-layer design,
network integration in the MAC layer
– Routing layer
• Optimal routing protocol features
–
–
–
–
Multiple performance metrics: how about hop-count?
Scalability
Robustness: link failures or congestion, load balancing
Efficient routing with mesh infra
Advances and research challenges
– Routing layer (con’t)
• Routing protocols with various performance metrics
– Link quality source routing (LQSR)
– Link quality metrics
– Expected transmission count (ETX)
• Multi-radio routing
– Weighted cumulative expected transmission time (WCETT)
– Link quality metric and the minimum hop-count
• Multi-path routing
– Better load balancing and high fault tolerance
• Hierarchical routing
– Clusters
Advances and research challenges
– Routing layer (con’t)
• Geographic routing
– Position information of nodes in the vicinity and the destination
node
– Delivery is not guaranteed even if a path exists between source and
destination
• Open research issues
–
–
–
–
Scalability
Better performance metrics
Routing/MAC cross-layer design
Efficient mesh routing
Advances and research challenges
– Transport layer
• Reliable data transport
– Non-congestion packet loss
– Unknown link failure due to wireless channels and mobility in
mesh clients
– Network asymmetry: TCP is critically dependent on ACK
– Large RTT variations
• Ad hoc transport protocol (ATP)
– Rate-based and quick-start
– Congestion detection is a delay-based approach -> congestion
cause
– No retransmission timeout
– Congestion control and reliability are decoupled
• Real-time delivery
– A rate control protocol (RCP) is needed to work with UDP
Advances and research challenges
– Transport layer (con’t)
• Open research issues
– Cross-layer solution to network asymmetry
– Adaptive TCP on various wireless networks
– Application layer
• Internet access
• Distributed information storage and sharing
• Information exchange across multiple wireless networks
– Improve existing applications layer protocols
– Propose new application-layer protocols for distributed information
sharing
– Develop innovative applications for WMNs
Advances and research challenges
– Network management
• Mobility management
– Distributed mobility management
• Power management
• Network monitoring
– Security
• Factors such as distributed network architecture, vulnerability of
channels and nodes in the shared wireless medium, and the
dynamic change of network topology
• An AAA centralized server?
• No central authroity for managing security keys
• An open issue: A distributed authentication and authorization
sheme with secure key management
Advances and research challenges
– Cross-layer design
• One approach is to improve the performance of a protocol layer
by taking into account parameters in other protocol layers.
• Another approach is to merge several protocols into one
component.