Connectivity Resotration Techniques for Wireless Sensor and Actor Network (WSAN), A Review

—Wireless Sensor and actor networks (WSANs) are the most promising research area in the field of wireless communication. It consists of large number of small independent sensor and powerful actor nodes equipped with communication and computation capabilities. Actors gather sensor’s data and react collaboratively to attain application particular assignments. A powerful connected inter-actor network is required to coordinate its operations. Actor node may fail due to the battery depletion or any hardware failure and this failure may divide the network into disjoint segments. This problem can degrade the network performance but also reduce the efficiency and effectiveness of the network. To restore the network into its original state, the researchers have proposed many connectivity restoration techniques during last few years. This paper provides a brief review of the existing connectivity restoration techniques for WSANs with their advantages and limitations.


INTRODUCTION
Wireless Sensor networks (WSNs) have gained a substantial attention in recent years.The wide spectrum of WSNs applications has open many research areas and it is being considered the most substantial modern technologies of recent times.As a result, huge research work has been done in different areas of WSNs.Wireless Sensor and Actor Network (WSAN) is a new paradigm that has reflected a significant impact on recent wireless sensor technologies.WSAN consists of sensor and actor nodes.Sensor and actor nodes in WSAN can be static or mobile.Sensors have limited power and energy resources and are responsible for sensing the important data from an area of interest and transmit it to the actor / base station.Actors are more powerful and high performance nodes furnished with more energy and computation resources, and work in close collaboration with the sensors to perform the fruitful tasks.Figure 1 shows the communication design of WSAN, in which sensors sense data from the environment and send to actors.Processing of data can be performed by actors or may send to base station for further processing.
Sensor and actor nodes can be failed due to battery depletion, hardware failure, or any external attack.As a result, network gets partitioned into disjoint segments, resulting in performance degradation [1].Therefore, connectivity restoration process is needed to be performed.Human intervention is very less in such applications and there is no observer, who observe the network and takes timely decision.Moreover, it is not an easy task to replace the faulty nodes with the new nodes, especially in a harsh environment like the battlefield or dense forest.So a comprehensive mechanism is required which replaces the failed node with any of its neighboring.Collaboration among the healthy actor nodes is required to restore the connectivity at the pre-failure level.Node's mobility can improve the overall performance of the network like connectivity, coverage and network lifetime [2].These recovery processes cause messaging overhead as well.Normally, these algorithms deal with single actor node failure, and do not consider the efficiency of the resources and also lack of emphasis on recovery time.The main objective of the review paper is to provide a brief review of the connectivity restoration techniques for wireless sensor and actor network.Moreover, this review paper also provides a comparison of different connectivity restoration techniques for WSAN with their objectives, advantages and limitations.
Rest of the paper is organized as follows.Section II provides the overview on need of study.Section III contains the background of the study, whereas Section IV consists of the review of the existing techniques and finally Section V concludes the review paper.

II. NEED OF STUDY
Wireless Sensor and Actor Network (WSAN) is implemented in different civilian and industrial applications.It has an exceptional architecture which differentiates it from the ordinary Wireless Sensor Network (WSN).Sensor Nodes have limited battery and power resources and are responsible for sensing and transmitting the sensed data to the actor nodes.Whereas, actors are more powerful and high performance nodes having high power and computational resources.Actor nodes have the ability to gather process and, send the collective and aggregated data to the base station.Actor nodes play a key role in WSAN.Therefore, special attention should be given to the actor nodes in order to increase the performance of the www.ijacsa.thesai.orgnetwork.Although, WSAN has a lot of benefits, but it is not free from obstacles and challenges.Communication of sensor to sensor, sensor to an actor, and actor to actor is very important.An efficient use and management of actor nodes can improve the overall performance of the network.The failure of an actor node causes disjoint segments in the network and sensed data could not reach to the base station.Therefore, connectivity restoration is the key problem to be addresses and solved.

III. BACKGROUND
In Wireless Sensor and Actor Networks (WSANs), sensor and actor nodes work together to perform certain tasks.Actor network is incorporated with the sensor network to make the wireless sensor and actor networks [3].WSANs can be affected due to the change of environment, any change in event detection, an actor mobility or failure of the actor due to the depletion of energy, attacks or any communication link issues.Failure of actors can divide the network into disjoint segments, and affect the whole network.Inter-actor nodes connectivity is significant [4].Due to the powerful characteristics of actor nodes, it can be managed through the relocation of mobile nodes but the actor failure can damage the network more than the ordinary sensor nodes.It can also affect the loss of coordination and connectivity between the nodes and restrict the event handling.For example, an actor node may be destroyed due to the catastrophic damage or enemy attack which disconnects the network and needs to be addressed instantaneously.Deployment of some relay nodes in the area can be a solution to replace the faulty nodes but it is not feasible in risky areas like war zones.Therefore, the early recovery process can be initiated through involving existing actor and sensor nodes by self-restoration techniques.Nodes repositioning methods have been introduced by number of researchers for restoring partitioned networks.
A typical WSAN architecture can be categorized into semiautomated and an automated architecture.These architectures are based on data passing and decision making.In an automated wireless sensor actor network, sensor nodes sense events and then send facts to their associated actor nodes which act as the base station.In semi-automated WSANs, sensor nodes sense facts from the field, and transfer to the sink node where sink node process the data, and communicate with actors to perform the necessary tasks, if needed.So coordination is very important in WSANs.
Fault tolerance techniques work in distributed manners.Fault tolerant techniques enable a system to perform its operations properly after single or multiple node failure.Figure 2 shows the node types in fault management techniques.Fault tolerant techniques are divided into critical and non-critical which solve the problem using 1-hop, 2-hop or multi-hop nodes information.Fault tolerance detection mechanism can be classified as proactive and reactive or hybrid.In proactive method, fault and restoration processes are addressed in the network setup in which redundant and backup nodes are deployed to ensure fault tolerance [5].In reactive techniques, it utilizes the in house resources of the network and performs the recovery process dynamically through repositioning of the nodes.It requires monitoring system to check the nodes status and recovery scope.Reactive techniques are divided into distributed or centralized and are discussed comprehensively in [6][7][8].Moreover, some techniques require a single node whereas some require a block of nodes to be moved to restore the connectivity of the network.Detailed taxonomy of such techniques is shown in the Figure 3.Some of these techniques will be discussed later in the paper.
The impact of a sensor/ an actor node failure can be different, according to the node type and its importance in the network.Fault management detection algorithms and restoration procedures can be categorized into critical and noncritical nodes [9][10][11].The failure of a critical node divides the network into disjoint segments.Most of the techniques describe the critical nodes by using 2-hop messages whereas some have the 1-hop message interchange to check the critical sensor/ actor nodes.A study was conducted using 1-hop to identify the critical actor nodes [12].To identify the critical node, the proposed technique calculates the distance of actor nodes from their adjacent nodes.If the distance is lesser than the neighbor's communication range, an actor will be considered as a non-critical node, otherwise critical.

Fig. 2. Fault Management
Coordination is the important factor in communication like sensor to sensor, sensor to actor and actor to actor.Network efficiency can be improved by communication between actors to actor.In WSAN, a sensor and actor node can be static or mobile.During the network lifetime, nodes move, so topology management is equally important.Nodes failure can divide the network into small segments and create the coverage hole in the sensing area as well.Topology management techniques can be performed automatically for fault management.

IV. REVIEW OF EXISTING TECHNIQUES
Actor node failure can divide the network of WSAN into segments.An actor may fail due to the fault in hardware, energy depletion, physical attacks or any communication link issues.Although, there are less chances of actor failure than sensor failure, but it can be controlled through relocation of some mobile nodes.Connectivity between the nodes and coordination will be lost in case of an actor failure, and leads to disjoint of the wireless sensor actor network.Fault tolerance is an ability of the network to do its work smoothly in response to node failure [13].
WSAN are mostly deployed in tough areas like the battlefield, dense forest or massive destruction areas and suppose to do work for the maximum period of time.Such networks are normally deployed in far areas from the main control Centre, so connectivity restoration in an efficient way is

2-hop
Multi-hop Non-Critical www.ijacsa.thesai.org a quite difficult process.Connectivity restoration process should be a distributed, self-healing and localized.Moreover, this process should be so fast that it reduces the impact of node failures, reduce overhead such that distance travelled, a number of messages while using limited energy supply.If a node travels too much to restore connectivity, then it will consume more energy and may also affect another network disconnection, especially if it is a cut-vertex.Moreover, it is difficult to find cut-vertex in large scale WSAN in a centralized and timely manner.So, connectivity restoration is a very challenging task in a distributed, localized, and an efficient manner.Reactive restoration techniques act passively and are initiated when a node failure occurs.No redundant resources are required in this case.Cooperative communication was introduced [14] which allowed a node to send a message away from its communication radius via its neighbor's.Two nodes can communicate with each other only if the received average Signal to Noise Ratio (SNR) is not less than the threshold.Strength of the signal reduces when we increase the transmission.Collaborative Single Node Failure Restoration algorithm (CSFR) [15] uses the cooperative communication approach to restore connectivity.It has low overhead but still has a long term process which consumes more energy and time.Therefore, it is not a suitable technique.In most of the reactive techniques, it reconnects a network by replacing a failure node with an appropriate backup node.It will be a recursive process that may relocate other nodes as well.It is mentioned earlier that only a cut-vertex node may break the network connectivity.A lot of techniques are available to detect whether the node is a cut vertex or not and then treat with its failure as well like NNN, DARA, PDARA, and PCR.In the improved version of PADRA, it forms a connecting dominating set (CDS).PADRA notifies in advance to a particular node about the partition occurrence in case of failure occurred.
Distributed Actor Recovery Algorithm (DARA) [16] finds a cut vertex by using two-hop neighborhood information.When a failure occurs, neighbors of the faulty node will select the most proper backup node.It will consider node's degree, distance and inform of its sibling nodes.Detailed process of identifying a cut vertex is not given in DARA technique in details.It is improved in PDARA in which it forms a connected dominating set (CDS).PDARA informs a particular node in advance whether a partition has occurred in case of failure.Nearest Non-Critical Neighbor (NNN) [17] decides whether it is cut-vertex or not and whether it preserves 2-hop neighbor's information of nodes.Distributed Partitioning Detection and Connectivity Restoration (DCR) algorithm [18] finds critical and non-critical actors in advance on the basis of information available locally and designates non-critical neighbor actors as backup actors.Once the failure of node occurs, the backup actor starts a recovery process which may include a coordinated method for relocation of multi actors.Recovery Algorithm for Multiple Nodes Failure (RAM) [18] handles two adjacent nodes failure simultaneously.It is a distributed hybrid technique which finds critical actors and assigns its backup nodes as well.Year-wise detailed comparison of different techniques is shown in Table 1.It covers techniques presented during the year 2007-2018 along with their objectives, node type, movement type, and node failure type.It shows that techniques are centralized or distributed and which technique has a capacity to restore connectivity of normal or cut-vertex node.This table also covers node movements like direct, cascade or block and also tells about the information that which technique has a capacity to restore connectivity of single, double or multi-node failure.
Recovery through inward Motion (RIM) [19] is a distributed technique which restores connectivity of critical and ordinary node.It maintains 1-hop information of each node; identifies nodes failure and starts recovery process by moving neighboring in cascade movement.All the one-hop neighboring nodes move towards the failed node till the distance is '' R C /2''.It ignores the impact of coverage after restoring connectivity and does not differentiate between the positions of nodes.RIM is very simple and an efficient technique, but its performance reduces in the dense networks and messages overhead is be very high.
Node Recovery through Active Spare designation (NORAS) [20] is a recovery algorithm which can identify critical nodes and finds the backup nodes for such critical nodes before the failure occurs.After that, these backup nodes participate to restore connectivity.It also takes care of coverage and connectivity in an integrated manner.NORAS stores 2-hop information which causes the large communication overhead in the network.Moreover, the movement of nodes is increased as compared to other algorithms.Least Movement Topology Repair Algorithm (LeMoToR) [23] is a localized and distributed technique which solves the network partitioning problem with less number of nodes movements and messaging overhead.It utilizes the path discovery activities at the time of connectivity restoration to identify the topology structure and takes suitable action accordingly.Faulty node is replaced with the neighboring node from the smallest disconnected block.This algorithm uses the recursive process to find the best route of recovery in which a lot of computation is required.There is no clear approach used to find the smallest block, either through depth first or greedy approach among the disconnected blocks.Moreover, communication overheads will be high while finding the smallest block.
Least Disruptive Topology Repair (LeDiR) [24] is another localized and distributed algorithm which can detect cut-vertex and execute recovery from node failure using path discovery and routing information.Neighboring nodes of the faulty node will recomputed their routing tables and develop enrolment decisions for the recovery process.Each node calculates the shortest path to other node and updates its information in the routing table.After the node failure, its one-hop neighbor's will check if the failed node is critical or not.Neighbor from the smallest block will move to replace the critical node.If more than one neighboring nodes are a part of the smallest block, then the nearest node from the faulty actor node will be selected to cope block movement.It consumes more energy because all the nodes of the block will participate in the recovery process.Moreover, smallest block calculation performed at the time of recovery which is another drawback of this algorithm.
Distributed Prioritized Connectivity Restoration Algorithm (DPCRA) [25] is used to restore connectivity and partitions by using a few numbers of nodes only.It identifies the negative effects of actors on partitions.The recovery process did locally while storing limited information in every node.Numerous backup nodes are used for the partition recovery in the network.This algorithm failed to address proper backup node selection criteria and as a result, there are more chances of failure of nodes.It can also affect the overall network performance and energy consumption as well.
Advanced-self-healing Connectivity Recovery Algorithm (ACRA) [26] defines the nature of the actor node that whether the failed node is cut-vertex or the connectivity of the nodes by depth first search technique.It will restore the connectivity of cut vertex node in which an actor node with more transmission power and high coverage area is selected to take part in the recovery process.Sensors and actor nodes are deployed randomly and form clusters.Every node has a system to detect failure of a normal node and also cut vertex actor nodes.When a cut vertex node fails, a neighboring cluster head (CHs) sends a recovery message to all the nearest nodes, towards the sink node, till it finds the next actor node or CH.A stable sensor CHs is selected as per GA based criteria among the neighboring nodes as a joining router for connecting partitioned network.This algorithm uses more energy because of the cluster heads.Senor nodes are involved in the recovery process on the basis of sensor resources, and there are more chances of nodes failure.Details of some important connectivity restoration techniques for wireless sensor and actor networks are presented in Table 2. Main objectives of some of vital techniques and their limitations summary are shown in Table 2. Limitations found in earlier algorithms have been addressed in the later techniques, whereas most of the techniques need researcher's attention to resolve these issues.
Hybrid Connectivity Restoration (HCR) algorithm [27] works proactively during selection process and reactively in motion phase.An actor selects the backup node through its 1hop neighboring table and notifies the backup node to oversee this process.When a node fails, its backup node try to move to that position in order to restore connectivity.It is a localized process which is repeated until connectivity is restored.HCR select the backup node which has to travel short.Moreover, HCR tries to reduce the number of messages by forwarding node failure information to its backup node only.It is an effective scheme with low complexity.It deals single node failure at a time, handles sequential nodes failure only and doesn't handle the coverage issues.
Efficient actor recovery paradigm (EAR) [28] is a recovery technique which can differentiate between critical and noncritical nodes.It allocates an appropriate backup node from its neighbor which is chosen on the basis of its signal strength, and control in its surroundings.It is supported by three algorithms.Node Monitoring and Critical Node Detection (NMCND) algorithm which monitors nodes and tells about the nodes type, and also handles the packet forwarding process if the primary node fails.Network integration and Message Forwarding (NIMF) is introduced to send packet.Process Based Routing for Node Failure Avoiding Algorithm (PRNFA) was developed to handle the routing process in which redundant messages was reduced to avoided network congestion.The main goal of this algorithm is to improve the node recovery process while maintaining the Quality of Services (QoS).
Distributed autonomous connectivity restoration method based on the finite state machine (DCRMF) [29] is a technique which looks for critical nodes in the region and defines how to reposition the related nodes.It performs restoration autonomously.Critical nodes updating process is launched after every restoration process.It relocates the non-dominating nodes from the neighboring nodes, so the total moving distance is reduced.It can effectively reduce the movement overheads of the nodes in connectivity restoration process.A distributed localized connectivity restoration algorithm is introduced to handle the multiple nodes failure problem.A few nodes will be relocated with less moving distances.
Permanent Relocation Algorithm for Centralized Actor Recovery (PRACAR) [30] replaces the failed actor by one of its neighbor permanently.It is a self-route recovery algorithm which finds the optimum path to the sensors.Actor nodes are relocated permanently to a new location.This permanent placement of the redundant node at the position of the faulty node will stop the extra movements.This is an energy efficient technique which saves energy to do other key tasks of the network.www.ijacsa.thesai.org

V. CONCLUSION
Maintaining internode connectivity in an efficient manner is very important in applications of wireless sensor and actor networks.Failure of actor nodes may disjoint the network and QoS of the network may degrade.There are many algorithms so far presented to restore connectivity of actor nodes, which were briefly discussed above.Some techniques use large number of nodes movement, more distance travel to restore connectivity, which consume more energy during the restoration process.A comprehensive review on state of the art connectivity restoration techniques is provided, in which different limitations and drawbacks in the schemes are discussed which may be addressed and may be helpful for the researches to develop new connectivity restoration techniques in the light of these guidelines.

Fig. 3 .
Fig. 3. Detailed taxonomy of techniques C 3 R-Coverage Conscious Connectivity Restoration [21] algorithm deals with the restoration of connectivity and coverage simultaneously.In this technique, failed node is replaced with the nearest node temporarily, and then it returns

Network Resotration Centralized Single Node Failure MCDS ORC Distributed Proactive Single Node Failure DARA PADARA NDRA Two Nodes Failure MPADRA Reactive Single Node Failure C2AM VCR LeDIR LeMoToR LDMR C3R PADRA Two Nodes Failure DCRA Multinodes Failure DORMS AuR CCRA Hybrid Single Node Failure RIM DCR Two nodes Failure RAM
. A neighboring node comes forward to replace it.It deals only with a single node failure, and does not support simultaneous nodes failure.Unnecessary movements of the nodes are reduced by introducing Energy Centric Optimized Algorithm (ECR) which consumes a lot of energy.It performs better than RIM and NN because it localizes the failure recovery and introduces few changes in the network topology.

TABLE II .
ACTOR FAILURE TECHNIQUES OBJECTIVES AND LIMITATIONS