Enhancement of qos in multihop wireless networks by delivering cbr using lb a...
AWN_FINAL_TERM_PAPER
1. Survey on Routing Algorithms for Anycast in Wireless Mesh Networks
Chandrasekar Hariharan (U00745288)
hariharan.11@wright.edu
Abstract:
Using Anycast based routing algorithms instead of
traditional routing algorithms, increases the
performance of the wireless mesh network. Selecting
an appropriate gateway over many available
gateways is the main problem discussed in this paper.
As this gateway selection problem is NP-Complete, a
heuristic based approach is used. Three popular
anycast based heuristics is applied for gateway
selection problem. First heuristic uses anycast based
algorithm to optimize load balancing parameters in a
distributed Wireless Mesh Network. Second heuristic
uses anycast based routing algorithm to improve the
quality of service and minimize interference of the
entire network. Third heuristic uses ant-colony based
heuristic to optimize the entire network. Traditional
protocol, AODV, is compared with all three anycast
implementations and best routing protocol to use in
wireless mesh network is determined. The proposed
survey indicates that adapting based approach, (i.e)
find the appropriate gateway adaptively, is best
suited than an approach which considers fixed route
from source to destination.
Keywords: Anycast, Pheromone, Ant-Colony,
GSQAR, MMAMBA, AGSA.
1. INTRODUCTION
Wireless mesh network is a form of wireless ad hoc
network which is made of radio nodes such as clients,
routers and gateways, typically well organized in a mesh
topology. As the usage of broadband internet has
substantially increased, the need for reliable, fast and
low cost networks has increased.
This tem paper is part of the course work of CEG-7470-Advanced Wireless
Networks, to be submitted on 12/10/2014. The instructor- Dr. Bin Wang
Author- Chandrasekar Hariharan, with graduate school in wright state
university, in the computer science department, 3640 colonel glenn hwy,
Dayton, oh 45324 USA (E-mail : Hariharan.11@wright.edu).
Wireless Mesh Network is one such network, which is
reliable and can communicate with rest of the nodes in
the network even if one node is not functional. The
important property of this network is that this network
can self-form and self-heal in-case of any problem. In
this network, multiple gateways are used to transfer data.
Selecting the appropriate gateway so as to increase the
total communication quality, balance network load and
minimize optimal cost is gateway selection problem.
An Anycast address selects one best server among group
of servers, thereby balancing the load and improving the
overall performance of this network. In Anycast
communications, the client needs no knowledge of the
location of server in obtaining an appropriate node
corresponding to a specific service.
Increasing the capacity of the network by selecting the
appropriate gateway node to communicate with
destination is an NP-Complete problem. Heuristics
based on Anycast routing protocol has been used to
solve this problem in an efficient way.
A gateway selection algorithm based on Anycast is
proposed so as to minimize the congestion occurring at
the gateways [1]. A Quality aware Anycast algorithm is
proposed for this problem, which is used for VOIP
communications, ensuring reliable transmission of data
over the network [2]. Multi-path Anycast Routing Based
on Ant Colony Optimization in Multi-gateway WMN
uses a heuristic called ant colony algorithm to preserve
both of the above mentioned properties [3].
Gateway selection algorithm is optimized using three
heuristics mentioned above. Each heuristic improves the
performance of the solution and results in relatively
better performance.
2. BACKGROUND
As all the clients share a common gateway to send data
across the network, the bandwidth of the network is
reduced. Due to the gateway bottleneck, the capacity of
2. the network flow is greatly reduced. Throughput per
client and reliability of the network is greatly reduced
because of congestion in the gateway network. Anycast
routing network selects the best node amongst group of
servers [1].
The network load balance and resource management is
efficient when using anycast protocol [2]. From the
above two facts, it is evident that use of Anycast in
solving gateway selection problem, gives optimal results
than any traditional routing protocols.
GSQAR (Gateway Selection and Quality Aware Anycast
Protocol for Wireless Mesh Networks), proposes a
heuristic which solves the gateway selection problem for
a fewer number of packets. It increases the throughput
relatively better than the any random gateway selection
scheme. GSQAR minimizes the interference, congestion
and maximizes the throughput.
As we use Anycast to solve the gateway selection
problem, the gateway selection problem is reduced to
Anycast routing problem. So far there is no precise
method to solve this problem in polynomial time [4].
Ant colony algorithm solves combinatory and
optimization problems by simulating the conduct of ants.
Pheromone, a substance left by the ant on looking for
food is the parameter used to find the shortest path to the
destination. This is a heuristic, which comes under the
class of evolutionary algorithms, finds the optimal
gateway adaptively by choosing the best combination.
Ant climbs its way all along global maxima and does not
converge in to local maxima.
3. RELATED WORK
The simplest way for solving the gateway selection
problem is by measuring the number of hops from
source to destination. The advantage of this method is
that we can send a data packet from source to destination
in the shortest path. The disadvantage of this method is
that the gateway chosen to send a packet will be
congested easily [1].
An adaptive gateway discovery method was proposed by
ZHANG which establishes a route to the new gateway if
it detects any node is going to leave the network. The
space and time overhead for implementing this
algorithm is large for larger networks. Multi-Gateway
wireless mesh routing protocol uses Dynamic Source
Routing protocol to find the cheapest gateway in the
network. The route which receives the first
acknowledgment amongst all the paths is considered to
be the cheapest and best path. This algorithm does not
work well with failure in the network [2].
A multi-gateway association scheme is proposed, where
the shortest path from source to destination is found on a
greedy manner. By selecting gateway in this manner, the
solution most likely converges to a local maximum path
than a global maximum [2].
Another dynamic programming approach is proposed to
reduce the space and time required to solve this problem.
By this method, each and every node memorizes its
nearest gateway node. All the data packets sent are to its
nearest gateway node. The advantages of this method are
that the computations performed to send the data packets
along the shortest path is less. The disadvantage of this
method is that the gateway to which all the clients send
its packet will be congested [2].
Anycast based networks solves the problem of self-
healing and self-forming in Delay/Disruption tolerant
networks. In this method, genetic algorithm based
implementation is used so as to choose a population
which evolves through several mutations or re-
combinations, eliminates the overhead due to
delay/disruption. This is an important property which
can be inherited for use in to Wireless Mesh Networks
[3].
An efficient dynamic multicast routing protocol which
uses anycast for routing uses off-tree anycast routing and
on-tree dynamic routing. This protocol uses shared tree
based approach, which combine them to form a virtual
anycast group. Delay is used as a parameter to route
amongst all the nodes to find the shortest path. In on-tree
dynamic routing, the shared trees are used dynamically,
to route amongst the nodes. To ensure reliability in the
network, backup nodes are used; in-case if a deadlock or
node used to communicate to the destination is dead
[12].
Multipath routing provides increased throughput and
performance. Multipath networks provide good load
balancing support and finds best path than single path
routing. Mobile-adhoc-networks use multipath routing to
communicate faster. In this paper, two on-demand
methods have been proposed to search for a node. In a
distributed environment, searching for a node amongst
all its peers to send data is difficult. A new metric is
proposed in this paper to reduce the cost and time
required to search for a node in a distributed system
[13].
4. PROBLEM STATEMENT
A Wireless Mesh Network which has multiple access
points, communicate with in the network, using Access
Points. Gateway Access Point is flooded heavily when
compared to all the other Access Points due to the nature
3. of routing. Due to this problem, the overall performance
of Wireless Mesh Networks is affected. Hence we
survey the Anycast routing protocols, which increases
the efficiency and reliability of the Wireless Mesh
Network. Once we adapt Anycast routing, the gateway
selection problem is typically reduced in to Anycast
routing Wireless Mesh Networks. Routing algorithms
for Multi-Path Anycast routing in Multi-gateway
Wireless mesh network are compared using end-to-end
delay and success in packet delivery.
5. DESCRIPTION OF EXISTING ALGORITHMS
5.1 Anycast-based Gateway Selection Algorithm for
Wireless Mesh Network
The gateway selection problem of wireless mesh
networks is modelled as an undirected graph. All the
mesh clients and mesh routers are modelled as vertices.
Nodes communicate amongst them through mesh
routers. Any packet which is sent to the internet goes
through gateway vertices. Neighbor nodes of a node can
be defined as the nodes which are adjacent to that node.
Basic graph problems such as finding the shortest path,
detecting the cycles in the graph are solved by default.
The classical gateway selection problem is modelled as
an anycast QoS request in a graph, initiated by the client,
with requested connection parameters such as hops,
delay etc.
Fig 1.1 End-to-End Average Delays
The nodes which are present in the network are
classified based on their features. Three types of nodes
are present in this Wireless Mesh Network. When a
unique node on the network, has its address range which
is similar to the address range of the anycast address
range, that node is called anycast group header.
Secondly, there are nodes in the network whose primary
operation is to carry the packets and receive the same
from the gateway. These nodes act as the backbone node
for the gateway and are called as supporting nodes.
Third type of nodes which typically provides internet
access to all the other nodes in the network is gateway
anycast group member. All the members of an anycast
group share common anycast address.
As gateway selection problem is NP-Complete, it is
impossible to solve the problem unless each and every
possible combination is evaluated. This paper proposes a
heuristic where selecting the best gateway depends upon
the minimum delay. This metric serves the purpose for
load balancing, reducing gateway congestion and
selecting a better route to the destination.
From the above graph it is evident that on comparison
with a traditional routing protocol, anycast is best suited
for Wireless Mesh Network. When the connection
numbers increase, the number of packets delivered is
greater in Anycast.
As we have modelled the problem as graph problem,
minimum spanning tree is used to select the shortest path
from any mesh client to gateway. Interference and delays
are two parameters which are used to optimize the
minimum spanning tree found. Kruskal’s algorithm
which starts selecting the nodes with minimum value to
determine the minimum spanning tree is used in this
context. Edges with minimum interference are added to
the tree at every instance to obtain the minimum shortest
path from source to gateway node. Once an anycast tree
is constructed, the minimum value from source to
gateway is found. Join group and leave group are two
operations performed when a gateway joins or leaves the
internet.
The disadvantage of this algorithm is that it converges to
local optimum not global maximum. When the network
is huge, minimum spanning tree, which works on a
greedy manner converges to its local optimum.
5.2 GSQAR: A Quality Aware Anycast Routing Protocol
for Wireless Mesh Networks
In this paper, a centralized gateway and route selection
scheme is proposed. With effect to improvise the total
quality of communication, this paper proposes a novel
approach by considering interference of packet
transmission.
Assigning list of source nodes to list of gateways in a
way which reduces the delay and increases the total
number of packets sent is the main theme of the
problem. Use of anycast for solving this problem with
Multi-gateway WMN’s is preferred for providing quality
of service.
4. This paper tries to develop a routing metric which
captures a successful multihop route. This is done using
two parameters such as probability of a successful
reception of a packet send and the time taken to send a
packet. The factors which contribute to delay on
transmission are analyzed and optimized across the
network.
To model the probability of success parameter, all
possible causes for a network’s failure is studied. Three
major reasons for failures are found. Interference from
nodes that are outside the reception range of the CTS
packet is one major reason which affects the failure of
the data transmitted. Secondly data packets fail due to
the interference which is present in the transmission
range. The third major reason for failure is interference
caused by the neighboring nodes.
Interference outside the reception range is caused either
due to RTS or data packets. As the length of RTS packet
when compared to the entire data packet is negligible,
only the data packet’s failure is considered. Interference
with in the transmission range occurs due to an
overlapping transmission from any nearby node Failure
of ACK packet can only be caused from any RTS packet
sent from neighbors. As this interference is very low
when compared to others, it can be neglected.
The queuing and access delay plays a major role in
affecting success of the data transmitted. Accessing
channel, back-off transmission and re-transmission of
data to many nodes near the neighborhood are some
critical factors which contributes to the delaying the
entire transmission.
Fig 1.2 End-to-End Average Delays:
A routing quality metric is proposed with the above facts
which assesses the entire network and tries to reduce the
delay and maximize the probability of success. For any
route the product of probability of success and its delay
gives link delay and the sum of all link delay gives total
delay.
For the above mentioned problem, in order to increase
the success rate, anycast based routing protocol is
proposed. This quality aware anycast is based on the
assumption that all the neighboring nodes are already
known. An RREQ packet is sent to the destination.
Based on its response, the corresponding path is
evaluated and stored. List of all stored paths to reach
destination is called candidate routes. The quality of
each and every route is recorded in a table called quality
table.
When a new node enters the network, this paper,
provides two techniques to obtain the shortest route. In
first technique, incremental method, where all nodes are
updated if and only if it has a less interference value
when compared with the previous route. Secondly there
is a global method where the route for all the sources is
re-calculated. Global solution takes time, but gives more
accurate route predictions than Incremental solution.
From the above chart it is clear that using GSQAR for
wireless mesh networks in routing increases the
performance of the system when compared to traditional
protocols. Probability of success is maximized and end-
to-end delay is minimized.
The disadvantage of this protocol is that, the first method
uses local comparisons on incremental update, which
might be good for one node but not for the entire
network. Global update, re-calculating each and every
networks path, upon addition of a new network is time
consuming.
Hence to solve the compute global insertions and
deletions of nodes, we will look at Multi-path Anycast
Routing, Based on ant colony optimization.
5.3 Multi-path Anycast Routing Based on Ant Colony
Optimization in Multi-gateway WMN
In this heuristic, we use anycast routing based on ant
colony optimization. Mesh Routers in wireless mesh
networks do not frequently change or move. Due to this
nature, local congestions on the wireless mesh networks
are increased. By increasing the number of gateway
nodes, selecting the appropriate gateway node to access
the internet is the main theme of this paper.
Ant colony algorithm is globally optimized and updates
its results based on the information provided by
pheromone. Performance of wireless mesh networks is
improved considerably on comparison with all
traditional routing protocols. MMAMBA (Multi-path
5. Anycast Routing Based on Ant Colony Optimization) is
compared AOMDV protocol to differentiate the
performance. Ants leave substance called pheromone, in
all the paths it leaves. The movement of ants is
determined on route which has high strength. This
positive feedback based mechanism is well suited and
preserves the property of self-healing and self-forming
wireless mesh networks.
MMAMBA chooses the best possible anycast group for
a node in the network to communicate with the internet.
The effective routes with largest probability are the route
selected. A new request packet called Fant is sent to
obtain the path travelled. If an intermediate node
receives this fant, it determines whether it is the best
node to reach the destination. If it is the best node, it
replies to the source. If it is not the best route, the
destination node and next hop information are reverse
back to the source.
The serial number of the fant is compared with the serial
numbers. If the fant serial number is greater, a new route
is preferred over the existing route. If the fant serial
number is lesser, the route which was obtained by
sending the RREQ packet is less optimal than the route
present in the routing table. If the fant serial number is
equal to the existing number, the algorithm checks if any
neighboring node is present in the route. If there are any
neighboring nodes, the routing table is updated with
them.
Fig 1.3 End-to-End Average Delays
If this fant packet reaches the end of the node or if fant
finds an effective route to the destination, a new bant
packet is sent to the source node. This bant packet
signifies that shortest path from a node to its gateway is
found and the node can connect to the gateway node.
Pheromone is updated based on the bant packet sent in
the reverse to source.
In this protocol, the delay and congestion information is
carried across the network, which constantly updates the
pheromone. Pheromone has the control as to choose the
node in the path or not.
The above table says that the use of anycast protocol in
gateway selection is much faster than the traditional
routing protocols. As the pheromone adapts to the
network if it finds a new path, addition of any newer
nodes does not impact much of the network. The
advantage of this approach is that once pheromone
learns entire network, it always gives the best route to
the destination.
6. COMPARISION
From the above existing algorithms, we compare use of
anycast routing protocol in contrast to traditional routing
protocols like AODV and AOMDV.
From Anycast-based Gateway Selection Algorithm for
Wireless Mesh Network, we can see that two protocols
perform well when the connection numbers are small.
When the connection numbers increases, AGSA
protocol performs well than traditional routing protocol.
From that fig 1.1, we can say that use of anycast routing
protocol in the wireless mesh networks is best suited
than any traditional protocol. This protocol is best used
for load balancing and reducing congestion in networks.
The disadvantage of this heuristic is that, it requires
entire nodes for processing when a node is inserted or
deleted. As the shortest path chosen is based on local
node’s smallest distance, adapting to global network
change is difficult.
.
Fig 1.4 End-to-End Delay Comparisons
6. Hence we use anycast networks in the next heuristic.
This Quality Aware Anycast Routing Protocol for
Wireless Mesh Networks provides a heuristic which
focuses on quality of communication. On comparison
with traditional AODV protocol, this protocol provides
better quality of service and reduces the delay much
larger than expected. The quality metric used in this
process is effective. The simulation results show that
GSQAR is effective in improving both the throughput
and delay performance in mutlihop environments in
comparison to traditional routing protocols.
The disadvantage of this GSQAR approach is that it
takes more time on adapting to addition of new nodes.
Even though this protocol performs better than
traditional protocols, it takes much time to compare with
each and every node in the network and obtain good
results.
Finally MMAMBA protocol is used to provide better
results for this problem. Ant-Colony optimization based
approach is used in selecting the appropriate gateway of
the network.
Many factors such as delay, hop counts, load of the
gateway and quality of the network are considered and
an optimal path to the destination node is found
adaptively. This protocol overcomes the disadvantages
of dynamic node insertions and deletions, as pheromones
monitor the entire network and gives optimal path
solution.
Based on the above graph, the end-to-end delay is very
less in MMAMBA network in comparison with GSQR
and AGSA. This indicates the collective performance
and optimal route selection is best chosen by
MMAMBA algorithm.
7. CONCLUSION
From the above results, it is evident that use of anycast
protocol is best preferred when compared to any
traditional routing protocols. Comparing three heuristics,
we find that MMAMBA protocol is best suited in any
Multi-path, Multi-gateway wireless mesh network. The
first two anycast routing algorithms proposed are best
suited for load balancing, ensuring quality of service.
Pheromone based anycast selection approach used in
MMAMBA, works well amongst all the heuristics
proposed for optimizing proposed NP-Complete
problem. This protocol solves congestion of single
gateway problem, ensures higher throughput and lower
end-to-end delay.
8. FUTURE WORK
Future work includes integrating multiple channels with
multiple radios to reduce co-channel interference. Using
this adaptive MMAMBA algorithm in MANET to send a
data packet in the best optimal route to improve the
overall performance of the system can be done. Swarm
Optimization or genetic algorithm based anycast
implementations can be used as the input for ant-colony
optimization approach. As the inputs routes provided by
the algorithms are genetically improvised, time taken to
find an optimal path, (i.e) number of comparisons to be
done in the system is greatly reduced.
9. REFERENCES
[1] Zhihui, G., Taoshen, L., & Xiaolan, Q. (2010,
August). Anycast-based gateway selection algorithm
for Wireless Mesh Network. In Computer Science
and Education (ICCSE), 2010 5th International
Conference on (pp. 1699-1702). IEEE.
[2] Pal, A., & Nasipuri, A. (2010, December). GSQAR:
A quality aware anycast routing protocol for wireless
mesh networks. In Global Telecommunications
Conference (GLOBECOM 2010), 2010 IEEE (pp. 1-
5). IEEE.
[3] Ling, S., Jie, C., & Xue-jun, Y. (2010, August).
Multi-path anycast routing based on ant colony
optimization in multi-gateway WMN. In Computer
Science and Education (ICCSE), 2010 5th
International Conference on (pp. 1694-1698). IEEE.
[4] Jianxin Wang, Yuan Zheng, Weijia Jia. An Anycast
Protocol based-ADOV in MANET. The 14th IEEE
International Symposiun on Personal, Indoor and
Mobile Radio Communication, 2003.
[5] Kassabaldlsi,El-Sharkaw I W A,Marks R
J.Swarm intelligence for routing in
communication networks[J].Global
Telecommunications,2001,6(6):3613-3617.
[6] Changui Shin, SungHo Kim, Sunshin An.Stable
gateway selection scheme based on MANET with
Internet[C]. The Sixth IEEE International Conference
on Computer and Information Technology,Dhaka,
Bangladesh, 2006, pp.80.
[7] ZHANG Kaijie, XIANG Yong, SHI Meilin. Multiple
gateways-based Internet connectivity for mobile ad
hoc networks [J]. JOURNAL OF TSINGHUA
UNIVERSITY (SCIENCE AND TECHNOLOGY.
2007, 47(1): 100-103.
[8] R.Draves, J.Padhye, B.Zill. Routing in multi-radio,
multi-hop wireless mesh networks[C]. ACM Annual
International Conference on Mobile Computing and
Nerworking, Philadelphia, PA, USA, September
2004, pp. 114-128.
[9] L. Song and Z. bing Xia, “An anycast routing
protocol for wireless mesh access network,” ICIE,
vol. 2, pp. 82–85, 2009.
[10] D. Nandiraju, L. Santhanam, N. Nandiraju, and D.
Agrawal, “Achieving load balancing in wireless mesh
7. networks through multiple gateways,” in IEEE
MASS, 2006, pp. 807–812.
[11] K. Sharif, L. Cao, Y. Wang, and T. A. Dahlberg, “A
hybrid anycast routing protocol for load balancing in
heterogeneous access networks,” in ICCCN, 2008,
pp. 99–104.
[12] Jia, W., Xu, G., Zhao, W., & Au, P. O. (2002).
Efficient internet multicast routing using anycast path
selection. Journal of Network and Systems
Management, 10(4), 417-438.
[13] Wu, K., & Harms, J. (2002). Multipath routing for
mobile ad hoc networks.Communications and
Networks, Journal of, 4(1), 48-58.
[14] Baumann, R., Heimlicher, S., Lenders, V., & May,
M. (2007, June). HEAT: Scalable routing in wireless
mesh networks using temperature fields. In World of
Wireless, Mobile and Multimedia Networks, 2007.
WoWMoM 2007. IEEE International Symposium on
a (pp. 1-9). IEEE.
[15] Rozner, E., Seshadri, J., Mehta, Y., & Qiu, L. (2006,
September). Simple opportunistic routing protocol
for wireless mesh networks. In Wireless Mesh
Networks, 2006. WiMesh 2006. 2nd IEEE Workshop
on (pp. 48-54). IEEE.
[16] Badia, L., Botta, A., & Lenzini, L. (2009). A genetic
approach to joint routing and link scheduling for
wireless mesh networks. Ad Hoc Networks, 7(4),
654-664.
[17] Akyildiz, I. F., Wang, X., & Wang, W. (2005).
Wireless mesh networks: a survey. Computer
networks, 47(4), 445-487.
[18] Alicherry, M., Bhatia, R., & Li, L. E. (2005, August).
Joint channel assignment and routing for throughput
optimization in multi-radio wireless mesh networks.
InProceedings of the 11th annual international
conference on Mobile computing and networking (pp.
58-72). ACM.