MULTI-CHANNEL WIRELESS

MESH NETWORKS

Advisor : Chia-Chi Huang

Co-Advisor : Chih-Min Yu
Student : Laith Alsmadi
2011/04/11
OUTLINE
 Introduction
 System Architecture

 Evaluation Metric

 A. Centralized Channel Assignment

 Input and Output
 Algorithm Overview
 B. Distributed Channel Assignment
 Load balancing routing
 Routing Metric
 Interface-Channel Assignment
 Simulation Results
 Conclusion
2
 References
INTRODUCTION
 A (WMN) wireless mesh network is similar in concept to
a mobile ad hoc network
 the IEEE 802.11b/g standards and IEEE 802.11a
standard provide 3 and 12 non-overlapped frequency
channels, respectively, which could be used
simultaneously within a neighborhood.
 Ability to utilize multiple channels substantially increase
the effective bandwidth available to wireless networks

3
SYSTEM ARCHITECTURE

4
SYSTEM ARCHITECTURE
 For direct communication, two nodes need to be within
communication/hearing range of each other, and need to have
a common channel assigned to them.
 Additionally, a pair of nodes using a same channel that is
within sense/interference range interferes with each other’s
communication, even if they cannot directly communicate.
 Node pairs using different channels can communicate
simultaneously without interference.
 The “virtual links” depict direct communication between
them; there are no physical links between them.

5
PRIMARY RESEARCH QUESTIONS

 Question 1: How to assign channels to interfaces?

 Question 2: How to route packets?

link loads

Routing Channel Assignment

link capacities
6
 Channel Assignment Problem
 Neighbor-to-interface binding: determines through which
interface a node uses to communicate with each of its
neighbors with whom it intends to establish a virtual link.
 Interface-to-channel binding: determines which radio channel
a network interface should use.
 Load Balancing Routing Problem
 Channel assignment depends on the load on each virtual
link , which in turns depends on routing

7
EVALUATION METRIC
 The goal of the channel assignment and routing is to
maximize to the overall network goodput, or the number
of bytes it can transport between the traffic aggregation
devices and the wired connectivity gateways within a
unit time.
 To formalize this goal, we define the cross-section
goodput of a network as

where C(a; gi) is the useful network bandwidth available between a traffic
aggregation device a and a gateway node gi. B(a) is the bandwidth
requirement between a traffic aggregation device a and the wired network. 8
A. CENTRALIZED CHANNEL
ASSIGNMENT INPUT AND OUTPUT
 The inputs to the combined channel assignment and
routing algorithm are

9
 (1) an estimated traffic load for all communicating node pairs
 (2) a wireless mesh network topology
 (3) the number of 802.11 network interfaces available on
each node and the number of non-overlapping radio channels.
 The outputs of this algorithm are
 (1) the channel bound to each 802.11 interface
 (2) the set of paths for every communicating node pairs in the
wireless mesh network.
ALGORITHM OVERVIEW
 An iterative algorithm
 Switches between channel assignment Traffic Profile
& routing Initial Link Load Estimation

 Starts the loop with initial link Initial Link Loads

Link Loads
loads Channel Assignment

 Terminates when Channels

Link Capacity Estimation

 Capacity ≥ load for all links; or
 Convergence observed Link Capacities

Routing
No proof of convergence or its speed
Link Loads

Capacity ≥ Load No
For All Links?

Yes
10
Channels
+ Routes
INITIAL LINK LOAD ESTIMATION
 the capacity of link l:
 where Q is the number of available channels, CQ is the capacity per
channel, and Ll are the number of virtual links within the interference
range of l.
 Based on these virtual link capacities, the routing algorithm
determines the initial routes and thus the initial link loads.
 we assume perfect load balancing across all acceptable paths
between each communicating node pair P(s,d), and the number of
acceptable paths that pass a link l Pl(s,d). Then the expected load on
link:
 Where B(s,d) is the estimated load between the node pair in the traffic
profile.

11
CHANNEL ASSIGNMENT
 The channel assignment problem

 Given: link loads

 Goal: assign channels . capacity ≥ load for all links
 NP-hard in general
 Uses greedy heuristics instead
GREEDY HEURISTICS
 Traverse links in
decreasing order of load load
100
 Link with high load gets
200
channel with less 150
250

contention 10

 Case 1 50

 Both A & B have unused

interfaces A
Channel 7
B
 Use channel with least
1 2
contention 6 7 3 interfaces
7

A’s channel list B’s channel list

13
Use Channel 7
TRAVERSING LINKS
 Case 2
 Only B has unused A
Channel 8
B
interface
2 1
 Use channel with least
3 8 3 interfaces
contention from A’s list 8

A’s channel list B’s channel list

 Case 3 Use Channel 8

 Neither node has unused A
Channel 5
B
interface
5
5 5
 Use common channel if
7 6 3 interfaces
there is one 9 8

A’s channel list B’s channel list 14

Use Channel 5
TRAVERSING LINKS
 Case 4
 No unused interface & no
common channel
Channel 2
5
 Merge one A’s channel
with one B’s channel
Channel 2
5
 Channels chosen to
minimize combined
Channel 2
contention A B
 Needs to merge channels 2
2 2
5
network-wide 3 6 3 interfaces
4 7

A’s channel list B’s channel list

Merge channel 2 & 5 15

Great disruption: channel 2 very busy; channel 5 silent

CONTENTION & LINK CAPACITY

 Level of contention

16
 Consider neighboring links within carrier sense range
using the same channel
 Sum up their loads

 Link capacity estimation

Load i
Capacityi   Capacity Channel
 Load j
jCSRange i  i, j: links in the same channel

Capacity of channel ill-defined

ROUTING
 two different routing algorithms
 (1) shortest path routing
 (2) randomized multipath routing.

 The shortest path refers to the shortest “feasible” path,

i.e., a path with sufficient available bandwidth and least
hop-count.
 The multi-path routing algorithm attempts to achieve
load balancing by distributing the traffic between a pair
of nodes among multiple available paths at run time.

17
B. DISTRIBUTED CHANNEL
ASSIGNMENT
LOAD BALANCING ROUTING
 Each WMN node needs to discover
a path to reach one or multiple
wired gateway node
 Each wired gateway node is the
root of a spanning tree, and each
WMN node attempt to participate
in one spanning tree
 The ADVERTISE packet sends out
contains the “cost” of reaching the
wired network

18
19
ROUTING METRIC
 Three different cost metrics.

 Hop count: minimum number of hops

 Gateway link capacity: residual capacity of the uplink that
connects the root gateway. Residual capacity of any link is
determined by subtracting the current usage of the link from
its overall capacity.
 Path capacity: minimum residual bandwidth of the path that
connects a WMN node to the wired network.

20
NEIGHBOR-INTERFACE BINDING
 Ripple effect

21
22
INTERFACE-CHANNEL ASSIGNMENT
 The channel assignment of a WMN node’s UP-NIC is
the responsibility of its parent.
 To assign channels to a WMN node’s DOWN-NIC
depend on that are least-used in its vicinity
 Each node periodically exchange its individual channel
usage information as a CHAL_USAGE packet with its
k+1 hop neighbor. where k is the ratio of the interference
range and the communication range.

23
VIRTUAL CONTROL NETWORK
 A WMN node need to communicate each control
message to its k hop physical neighbor
 A control message can be delivered through one or
multiple hops
 For efficiency reason, the broadcast control messages are
delivered using IP multicast

24
FAILURE RECOVERY

25
SIMULATION RESULTS
 As more channels are made available, the channel assignment algorithm
uses them to increase the overall network throughput.
 distributed channel assignment improves the network throughput 6 to 7
times as compared with a single-channel network.

27
 The network goodput increases with the number of gateway nodes in the
network

28
CONCLUSION
 The bandwidth issue is most severe for multi-hop
wireless mesh networks due to interference among
successive hops of an individual path as well as among
neighboring paths.
 A channel assignment needs to balance between
maintaining network connectivity and increasing
aggregate bandwidth
 The distributed channel assignment / routing algorithm
we developed for the WMN can achieve a factor of 6 to
7 throughput improvement compared to single channel
WMN
29
REFERENCES
 A. Raniwala, K. Gopalan, T. Chiueh; “Centralized
Channel Assignment and Routing Algorithms for Multi-
channel Wireless Mesh Networks”, ACM Mobile
Computing & Comm Review (MC2R), April '04.
 A. Raniwala and T. Chiueh, "Architecture and
algorithms for an IEEE 802.11-based multi-channel
wireless mesh network", in Proc. INFOCOM, 2005,
pp.2223-2234.

30