IAETSD JOURNAL FOR ADVANCED RESEARCH IN APPLIED SCIENCES ISSN NO: 2394-8442

Topology Changes in WSAN and Fault Tolerance

Pankaj R. Patil
Department of Computer Engineering, R.C. Patel Institute of Technology, Shirpur, North Maharashtra University
patil.pankaj01@gmail.com

Abstract
Wireless sensor and actor networks employ actor nodes within the wireless sensor network (WSN) which can process the sensed data
and perform certain actions. The employed actors should form and maintain a connected inter-actor network at the times. WSANs
often operate in harsh environments where actors can easily fail or may get damaged. This kind of failures can partition the inter-actor
network and thus eventually make the network useless. In order to handle such failures, one of the effective recovery methodologies is
to autonomously reposition a subset of the sensor nodes to restore the connectivity. Generally the Contemporary recovery schemes
either impose high node relocation overhead or extend some of the inter-actor data paths. The overview of such kinds of different
fault tolerance algorithms is provided here.

KeywordsFault Tolerance, Wireless Sensor Actor Network (WSAN), Network Recovery, Topology Management.

I. INTRODUCTION
Sensors are inexpensive and highly constrained in energy and processing capacity and actors are more capable nodes with relatively
more on board energy supply and richer computation and communication resources. Recent years have witnessed a growing interest in
the applications of wireless sensor-actor networks (WSANs) of particular interest are applications in remote and harsh areas in which
human intervention is risky or impractical .A WSAN consists of a set of miniaturized low-cost sensors that are spread in an area of
interest to measure ambient conditions in the vicinity. Actors usually coordinate their motion so that they stay reachable to each other.
However, a failure of an actor may cause the network to partition into disjoint blocks and would thus violate such a connectivity
requirement. The best recovery option in addition, tolerance of node failure cannot be orchestrated through a centralized scheme given
the autonomous operation of the network. On the other hand, distributed recovery will be very challenging since nodes in separate
partitions will not be able to reach each other to coordinate the recovery process. Therefore, contemporary schemes found in the
literature require every node to maintain partial knowledge of the network state. To avoid the excessive state-update overhead and to
expedite the connectivity restoration process, prior work relies on maintaining one- or two-hop neighbour lists and predetermines
some criteria for the nodes involvement in the recovery .However, one-hop-based schemes often impose high node repositioning
overhead, and the repaired inter-actor topology using two-hop schemes may differ significantly from its prefigure status.

II. LITERATURE SURVEY

A number of schemes have recently been proposed for restoring network connectivity in partitioned WSANs. All of these schemes
have mainly categorized into two groups as follows:

A) Recovery through Node Repositioning

B) Recovery by placement of Relay Nodes

A. Recovery through Node Repositioning

The main idea is to reposition some of the healthy nodes in the network to reinstate strong connectivity. For example,

a) Least-Disruptive topology Repair (LeDiR) Algorithm,

b) Distributed Actor Recovery Algorithm (DARA),
c) PArtition Detection and Recovery Algorithm (PADRA),
d) Recovery through Inward Motion (RIM)
e) Least Distance Movement Recovery (LDMR).

VOLUME 4, ISSUE 6, NOV/2017 72 http://iaetsdjaras.org/

IAETSD JOURNAL FOR ADVANCED RESEARCH IN APPLIED SCIENCES ISSN NO: 2394-8442

1. Least Disruptive topology Repair (LeDiR) Algorithm

LeDiR generally designed for recover from a single node failure. The node failure may cause in conflicting conditions for
LeDiR to converge successfully. A novel Least-Disruptive topology Repair (LeDiR) algorithm is proposed. For multiple nodes failure
at the same time, LeDiR tends to shrink the smallest block inward toward the failed node; it may negatively affect the node coverage.
Care for the path length between nodes [1]. LeDiR strives to restore network connectivity while minimizing the recovery overhead and
maintaining the shortest path lengths at their prefailure value. We can group the results into two sets:

1) Overhead related metrics and

2) Path length validation metrics.

When a node fails, its neighbours will individually consult their possibly incomplete routing table to decide on the appropriate
course of actions and define their role in the recovery if any. If the failed node is critical to the network connectivity, i.e., a node whose
failure causes the network to partition into disjoint blocks, the neighbour that belongs to the smallest block reacts.

Advantage:

The advantage of LeDiR is it reduces prefailure communication overhead that provided at the expense of overreacting to
failure of uncritical nodes.

Disadvantage:

In sparse topologies LeDiR does not appear to have advantage over RIM, RIM does not prevent the paths from being
extended, as shown later in this paper. The impact of the actors failure on the network topology can be very limited.

2. PArtition Detection and Recovery Algorithm (PADRA)

PArtition Detection and Recovery Algorithm (PADRA) require every node to maintain a list of their two-hop neighbors and
determine the scope of the recovery by checking whether the failed node is a cut vertex. A distributed PArtition Detection and
Recovery Algorithm (PADRA) is proposed [2]. Distributed PArtition Detection and Recovery Algorithm (PADRA) can find possible
partitioning in advance and restore the connectivity and minimized node movement and message overhead.

Advantage:

All the actors will move the same distance with this approach, it will significantly increase the TMA, since all the nodes in
that partition will be moving. So propose a hybrid solution, which will combine the both approaches.

Disadvantage:

Failure of an actor may cause the loss of multiple inter-actor communication links, partition the network if alternate paths
among the affected actors are not available, and stop the actuation capabilities of the actor.

3. Distributed Actor Recovery Algorithm (DARA):

DARA pursues a probabilistic scheme to identify cut vertices. A best candidate (BC) is selected from the one-hop neighbours
of the dead actor as a recovery initiator and to replace the faulty node. The BC selection criterion is based on the least node degree and
physical proximity to the faulty node [3]. The relocation procedure is recursively applied to handle any disconnected children. In other
words, cascaded movement is used to sustain network connectivity. It does not care for the path length between nodes. DARA
identifies a connected dominating set to determine a dominate node. The dominate does not directly move to the location of the node
failed. A cascaded motion is pursued to share the burden. It does not care for the path length between nodes. The main optimization
objective of DARA is to minimize the total distance travelled by the involved actors in order to limit the overhead by the movement.
DARA strives to minimize the messaging costs and maintain scalability and entire recovery process is distributed and enabling the
network to the self-heal without external supervision. DARA generally are in two ways describes for WSANs, such as first one is
DARA-1C and other is DARA-2C, are developed to address 1 and 2-connectivity. The goal of DARA-1C is to restore the network
connectivity by exploiting the mobility of actors nodes.

VOLUME 4, ISSUE 6, NOV/2017 73 http://iaetsdjaras.org/

IAETSD JOURNAL FOR ADVANCED RESEARCH IN APPLIED SCIENCES ISSN NO: 2394-8442

The basic concept is to pick one of the neighbours of a failed node in order to replace it. DARA-1C strives to pick the
neighbouring actor and that will trigger the least number of cascaded movements. The total distance travelled by actors is minimized.
DARA-2C strives to restore. The effect of an actor failure is analysed and shown to depend on whether the actor is a boundary node
or not. The failure of a boundary node may risk the biconnectivity of the network that sufficient condition is derived to the recovery
process [3] .Two versions of DARA-2C are proposed for which a decrease in the algorithm complexity is traded off for a potential
increase in the total distance that the actors collectively travel both DARA-1C and DAR A-2C are completely distributed and require
only the knowledge of two hop neighbours.

Advantage:

This is particularly very advantageous for DARA since the network would be partitioned into a very large block and few small blocks.

Disadvantage:

DARA algorithm picks the neighbour with the least degree to limit the scope of relocation. . It is also interesting to note that
local flooding in DARA-1CLF has not brought any advantages as the network is randomly deployed.

4. Recovery through Inward Motion (RIM) [4]:

In RIM, any lost link during the recovery will be re-established through cascaded relocation. The collective effect seems like
the network topology is shrinking inward. It reduces prefailure communication overhead [4]. Recovery through Inward Motion (RIM),
a distributed algorithm is proposed to efficiently restore network connectivity after a node failure. RIM is a distributed algorithm for
Recovery through Inward Motion. RIM restores the connectivity of a WSN through the efficient repositioning of some of its nodes.
RIM is a localized scheme that limits the scope of the recovery process. The main idea is that if a node fails, then its neighbours move
inward toward its position .The rationale is that these neighbours are the ones directly impacted by the failure. They can reach to each
other again, and the network connectivity is restored to its pre-failure status. The relocation procedure is recursively applied to handle
any nodes that get disconnected when one of their neighbours moves. RIM is simple and effective. It employs a simple procedure that
recovers from both serious and non-serious breaks in connectivity, without checking to see if the failed node is a cut vertex. The
recovery process is distributed and enabling the network to heal itself without external supervision.

Advantage:

RIM restores the connectivity of a WSN through the efficient repositioning of some of its nodes thus provides less overhead.

Disadvantage:

The network may get partitioned into multiple disjoint blocks and failures may affect both network coverage and connectivity,
this paper focuses on maintaining the latter when a node is lost.

5. Least Distance Movement Recovery (LDMR)

LDMR is a distributed recovery algorithm that exploits non cut-vertices in order to require the least travel distance from the
engaged nodes. The recovery process starts with the search phase where each neighbour broadcasts a message containing several
entries such as failed node ID, neighbour node ID and, Time-To-Live (TTL). Each neighbour chooses the best candidate among
received responses based on a certain criteria (e.g. distance). The picked candidates replace the moved nodes without additional node
relocation overhead. It compare proposed approach with RIM algorithm which is depends on cascaded movements. The Extensive
simulation experiments are carried out to validate the performance of node. It show that approach outperforms RIM for larger and
sparse networks. While the improvement in total travelled distance is achieved in these cases, loss of coverage after recovery operation
is also comparable [5]. LDMR avoids the cascaded relocation by sending messages to find the replacement for the neighbours of
Failure node after they move.

Advantage:

An adaptive approach can be devised combing the advantages of RIM and LDMR in order to perform well for all different
network sizes. Considering the coverage loss, the proposed approach has shown to yield comparable performance to RIM.

VOLUME 4, ISSUE 6, NOV/2017 74 http://iaetsdjaras.org/

IAETSD JOURNAL FOR ADVANCED RESEARCH IN APPLIED SCIENCES ISSN NO: 2394-8442

Disadvantage:

If number of nodes increases when the network becomes larger, the total distance traveled under RIM increases while it
decreases in the case of the LDMR.

B. Recovery by Placement of Relay Nodes

Distributed algorithm for Optimized Relay node placement using Minimum Steiner tree (DORMS) strive to minimize the
required number of relays nodes [6]. The spider Web Heuristic algorithm is proposed in this paper. Recovery from multiple
simultaneous failures in wireless sensor networks using minimum Steiner tree. Basically, both schemes try to avoid the introduction of
cut vertices so that some level of robustness, i.e., load balancing and high node degree, is introduced in the repaired network topology.
Spider Web and DORMS also strive to minimize the required number of relays. Both Spider Web and DORMS deploy relays inwards
toward the center of the deployment area[6].The former considers the segments situated at the perimeter and establishes a topology
that resembles a spider web. DORMS initially forms a star topology with all segments connected through a relay placed at the center of
the area. Then, adjacent branches are further optimized by forming a Steiner tree for connecting two segments and the center node to
reduce the required relay count [6].

Advantage:

The simulation results confirm the performance advantages of the Spider Web algorithm in terms of the required RN count
and the quality of the formed topology, which is assessed by the average node degree, expected path length, and node coverage. The
performance advantage for 1CSpiderWeb is quite significant. Therefore, 1C-SpiderWeb produces topologies with balanced traffic and
reduced latency.

Disadvantage:

Coordinated repositioning of nodes will not be feasible, because the network connectivity is severed such that the scope of
the damage cannot be determined, and several nodes will not reach other nodes.

III. CONCLUSIONS
In recent years, WSANs have growing attention due to their potential in many real life applications. In WSANs, reestablishing network
connectivity after node failure without extending the length of data paths is serious problem. To deal with this problem several
schemes are there. An overview of such schemes is provided in this paper. It is found that LeDiR algorithm is the only scheme that
restores connectivity by careful repositioning of nodes so that optimal changes in the topology are made. Most of the fault tolerance
schemes consider only single node failure considerations, but in real time scenario multiple node failure also occurs. Thus schemes
based on multiple node failure considerations are needed for effective fault tolerance.

REFERENCES
[1] Ameer, Anabasis, Mohammed Youngish and Uthman Barodi Recovering from Node Failure in Wireless Sensor-Actor Networks with Minimal
Topology Changes, IEEE Transaction On Vehicular Technology, page no 256-271, Vol.62 Jan2013.
[2] A. Thimmapuram, K. Akkaya, F. Senel, and S. Uludag, Distributed recovery of actor failures in wireless sensor and actor networks, in Proc.
IEEE WCNC, Las Vegas, NV, Mar. 2008, pp. 24802485.
[3] M. Younis, and K. Akkaya Movement-assisted connectivity restoration in wireless sensor and actor networks, IEEE Transaction on. Parallel
Distributed. System. vol. 20, no. 9, pp. 13661379, Sep. 2009.
[4] M. Younis, S. Lee, S. Gupta, and K. Fisher, A localized self-healing algorithm for networks of moveable sensor nodes,IEEE GLOBECOM,
New Orleans, November 2008.
[5] A. Alfadhly, and M. Younis, Least distance movement recovery approach for large scale wireless sensor-actor networks, in Proc.Workshop FedSenS,
Istanbul, Turkey, Jul. 2011.
[6] K.Akkaya and M. Younis, Bio-inspired relay node placement heuristics for repairing damaged Wireless sensor networks, IEEE Transaction in
May 2011.

VOLUME 4, ISSUE 6, NOV/2017 75 http://iaetsdjaras.org/