A cluster based minimum battery cost AODV routing using multipath route for zigbee

A Cluster Based Minimum Battery Cost AODVRouting Using Multipath Route for ZigBee

Ashutosh Bhatia Praveen Kaushik

Samsung , Software Operation, Department of CSE Bangalore, India MANIT, Bhopal, India

[email protected] [email protected]

Abstract— IEEE 802.15.4 standard is uniquely designed for low data rate wireless personal area networks (LR-WPANs). The IEEE 802.15.4 targets the applications such as industrial, agricultural, vehicular, residential, medical sensors and actuators which have more relaxed throughput requirements. ZigBee is a wireless technology based on IEEE 802.15.4. ZigBee routing uses Ad hoc On-demand Distance Vector (AODV) routing protocol. In this paper we present an improved version of AODV called Multipath Energy Aware AODV routing (ME-AODV), which utilizes the topology of network to divide it into one or more logical clusters and restricts the flooding of route request outside the cluster. The mesh links created at the time of cluster formation are used to decrease the routing path. ME-AODV uses nodes of the same cluster to share routing information, which significantly reduces the route path discovery. Since ZigBee routing is based on shortest-hop count, which causes overuse of a small set of nodes hence decreasing node as well as network lifetime. We also propose a mix of Ad hoc On-demand Multipath Distance Vector routing (AOMDV) and Minimal-Battery Cost Routing (MBCR) as an extension to AODV to increase the lifetime of network. The simulations have been performed using IEEE 802.15.4, ns-2 module.

Index Terms— ZigBee, IEEE 802.15.4, Ad hoc On-demand Distance Vector (AODV), Multipath Energy Aware AODV (MEAODV),Minimum Battery Cost Routing (MBCR).

I. INTRODUCTION In recent years, wireless communication has experienced

exponential growth caused by the need for connectivity. Wireless networking has followed a similar trend due to the increasing exchange of data in services such as the Internet, e-mail and data file transfer. The capabilities needed to deliver such services are characterized by an increasing need for data throughput. Other applications in fields such as industrial, agricultural, vehicular, residential, medical sensors and actuators have more relaxed throughput requirements. Moreover, these applications require substantially lower power consumption than is currently provided in existing standard implementations. For instance, battery-powered devices for certain types of industrial and medical sensors, smart tags and badges should last from several months to many years. A study [13] discusses the power requirement of such devices.

The IEEE standard, 802.15.4[1], defines the physical layer (PHY) and medium access control sublayer (MAC) specifications for low data rate wireless connectivity among relatively simple devices that consume minimal power and

typically operate in the Personal Operating Space (POS) of 10 meters or less. An 802.15.4 network can simply be a one-hop star, or, when lines of communication exceed 10 meters, a self-configuring, multi-hop network. ZigBee is a wireless technology that builds upon IEEE 802.15.4 standard. The ZigBee routing protocol uses the concept of Ad hoc On-demand Distance Vector (AODV) [10], ZigBee focuses on low data rate. However, routing overhead for route discovery may often interfere with network traffic especially for dense networks. The studies [5], [6] of wireless network performance on multi-hop show that interference on multi-hop networking influence scalability. It decreases network performances such as end-to-end delay and packet delivery ratio. Consequently, it is necessary to have routing protocols with reduced routing overhead. In this paper we present an improved version of AODV called Multipath Energy Aware AODV (ME-AODV), which utilizes the topology of ZigBee network to divide it into one or more logical clusters and restricts the flooding of route request messages outside the cluster. The other issue addressed in this work is related to the lifetime of the network. Since ZigBee routing is based on shortest-hop count which minimizes overall energy consumption of network, a small set of nodes get overused and the battery of those nodes may be drained in a short period of time, leading to potential network partition. We have proposed a mix of Minimum Battery Cost Routing (MBCR) [7] and Ad Hoc On-Demand Multipath Distance vector routing (AOMDV) [8] to increase the lifetime of ZigBee network. The rest of the paper is organized as follows. In section 2, we review the existing ZigBee routing and AODV protocol related to this work. In section 3, we discuss improved version of AODV called ME-AODV proposed by us. In section 4 we present experimental results and we'll conclude in section 5.

II. OVERVIEW OF ZIGBEE ZigBee technology is a low data rate, low power consuming, low cost; wireless networking protocol targeted towards automation, remote control and sensor network applications. The overall ZigBee architecture is composed of a set of layers. Each layer provides a defined set of services for the upper layer. In a ZigBee mesh topology, all nodes with

978-1-4244-3805-1/08/$25.00 ⓒ2008IEEE ICON 2008

routing capacity can communicate as peer-to-peer in a multi-hop mobile environment.

A. ZigBee Network Layer ZigBee network layer is a critical component that supports

ZigBee system like a skeleton. All ZigBee devices contain a network layer, but the network layer functionality that a device performs depends upon its role as a ZigBee Coordinator (ZC), ZigBee Router (ZR) or ZigBee End Device (ZED). All network functions can be performed by the ZC while reduced functionality is available to ZR and ZED.

The address allocation algorithm used in ZigBee allows any device to quickly identify whether a particular device belongs to a descendant of that device or elsewhere in the device tree as shown in Fig.1. As a result any device could make simple routing decision by sending a packet up or down the device tree. . In this case a packet is generated by the device 071f and is destined for 0002. The most significant advantages of tree routing are its simplicity and efficient use of resources.

Fig. 1. Tree Routing

B. ZigBee Network Formation A device that is capable of being a coordinator identifies a

channel that is relatively free of interference through energy scans and then works as the network's Personal Area Network (PAN) coordinator. Subsequently, other devices join the network by associating themselves to the PAN coordinator either as ZED or as ZR (the device capable of associating further children). When the PAN coordinator establishes a new network, it will assign itself a network address 0 and a network depth of Depth0 = 0. If node i wants to join the network and associate with node k, node k will become the father node of node i. And then according to its own network address Ak and network depth Depthk , node k will assign a network address Ai and a network depth of Depthi = Depthk + 1 to node i4. Network depth means the number of hops from a node to the PAN Coordinator. The parameter nwkMaxChildren (CM) represents the largest number of child nodes which can associate with a ZR or a ZC. The parameter nwkMaxRouters (RM) is the number of child nodes which can act as a ZR. The nwkMaxDepth (LM) means the maximum depth that a device can hold. A new node n is a Reduced Function Device (RFD), which means it has no routing ability. It is associated with a coordinator as its nth child node. According to its own depth d,

parent node k will assign child node n the network address An = Ak + Cskip(d) * Rm + n. If the new node is a Full Function Device (FFD), which means it has routing ability, parent node k will assign the network address such as An = Ak + 1 + Cskip(d) * (n-1). Every parent node will calculate the size of address sub block of its child nodes according to the function.

⎪⎩

⎪⎨⎧

−−−+

=−−∗+= −−

otherwiseRm

RmCRCifRdLC

dCskip dLmmm

mmmm

,1

*11),1(1

)( 1

Where d is the depth of parent node

C. ZigBee Routing The ZigBee uses a mix of Tree and AODV routing. In Tree

Routing, router forwards a data packet to the destination node whose network address is D. The network address and network depth of this router are equal to A and d respectively. It will first ascertain whether the destination node is its child node using expression A < D < A + Cskip(d-1). If the destination is its child node, the address of next hop node is:

⎪⎩

⎪⎨⎧

×⎥⎦

⎤⎢⎣

⎡ +−++= otherwisedCskip

dCskipADA

deviceendifDN ),(

)()1(1

,

If the designation is not a child node the next hop is father of this router. Although tree routing is simple and uses very less resource, it can be quite inefficient. Depending on how the network forms, two devices, in close proximity could join the network on remote branches of that network. Due to this the packet transfer between these two devices passes through many hops, whereas a single hop-path transmission between them could also be possible. As shown in Fig. 1.The packet flow from device ’071f’ to device ’0002’ was transmitted up and down the tree (through ’071e’, ’0000’ and ’0001’) while packet could be directly routed from ‘071f to device ‘0002’. To overcome the problem of Tree routing, ZigBee routing protocol adopts the well-studied public domain algorithm AODV. AODV is a pure on-demand route acquisition algorithm. When a node desires to send data to another node not having routing information, it first buffers the data packet and then initiates a route discovery process to locate a shortest path for the destination. The source node broadcasts a route request (RREQ) packet to its neighbors and then intermediate nodes receiving the RREQ rebroadcast it to their neighbors until it arrives at the destination node. AODV protocol implementation has been discussed in [14].

III. MULTIPATH ENERGY AWARE AODV ME-AODV is an improvement over AODV routing. The

ME-AODV uses two distinct ideas to improve the performance of AODV in ZigBee network. The first is the use

of cluster based approach, which has been inspired by earlier proposed work in cluster, based routing mechanism for mobile ad hoc networks [9]. Secondly, a mix of Ad hoc On-demand Multipath Distance Vector (AOMDV) [8] routing and Minimal-Battery Cost Routing (MBCR) [7] has been incorporated in existing ZigBee routing protocol. ME-AODV utilizes ZigBee network's cluster-tree topology to divide the network into logical clusters and assigns each logical cluster a unique cluster id. This cluster id along with the network address is used to avoid the flooding of route request messages during route discovery. The mesh links created at the time of cluster formation are used to decrease the routing path. In order to increase the lifetime of ZigBee network, ME-AODV employs a mix of AOMDV and MBCR. The ME-AODV maintains more than one path for a particular destination and uses them in a round robin fashion to evenly distribute the load between network devices. The simulations have been carried out using IEEE 802.15.4 ns-2 module [12]. Fig. 2 shows one of our simulated ZigBee Networks.

Fig. 2. An Example of ZigBee Network

A. Clustering in ME-AODV ME-AODV divides the network into logical clusters. A

cluster consists of several nodes with different roles such as a cluster header, gateway and cluster members. In order to form a cluster, ME-AODV defines the following rules.

1. Only ZC and ZR have the qualification of becoming a cluster header.

2. By default, end devices would be the member of their parent's cluster.

Fig. 3(a)-(f) illustrate the complete construction of clustering in our example ZigBee network. The clustering process takes the following steps. Fig. 3(a) shows an example of ZigBee network to demonstrate the clustering procedure. 1. The first cluster, cluster 0, is constructed with PAN

coordinator as cluster header. The PAN coordinator starts searching for other nodes within its radio range. The cluster 0 consists of PAN coordinator and found devices.

The coordinator then broadcasts a notification command to all found devices of cluster 0. If found node is not the child of coordinator, a mesh link is created between them. For example, as shown in Fig. 3(b), nodes 1, 2, 3, 4 are the found nodes and cluster 0 contains 0, 1, 2, 3, and 4. A mesh link (dotted line) is created between node 0 (PAN Coordinator) and node 16.

2. All the nodes of cluster 0 found in step 1 above (say Current), will start searching the existence of new nodes in their personal operating space. If the found node is a leaf node of cluster 0, then these nodes (Current and found node) are connected with a mesh-link, and the sibling relationship between them is memorized. If the found node is new node, then the new cluster is created and the found node is designated as new cluster header. If the found node is not the child of corresponding Current node then a mesh-link is created between them. For example, as shown in Fig. 3(c), nodes 8, 7 and 5 are newly found nodes and mesh links are created between 4-3 and 2-3 respectively. The current node will become the gateway for this newly created cluster. As shown in Fig. 3(d) nodes 4, 3 and 2 will become the gateway for these newly created clusters having 8, 7 and 2 as their cluster headers respectively.

3. All the cluster heads found in the step 2 (say Current), start searching for other nodes within their radio range. Let us take example of node 7 as Current node. There may be two possible situations:

a) The newly found node is already a cluster head. If the new node has parent-child relationship with Current, then only the parent node will have the qualification of becoming cluster head. If there is no relationship between them, then the node having higher degree will become the cluster head. In case of tie, the node with less numeric value of network address will become cluster head. In our case the degree of Current is greater than the degree of node 8, hence only Current will be the header of the cluster and node 8 will leave the nomination of cluster header and become leaf member of cluster. Fig. 3(e) illustrates this situation.

b) If the newly found node is not a member of any cluster then it will become a member of the cluster to which Current belongs. As shown in Fig. 3(e), 10 and 11 will become leaf members of the cluster with Current as the header. If found node is not the child of Current then a mesh link will be created between Current and found node. For example, as shown in Fig. 3(e), a mesh link is created between Current and 11.

4. The leaf nodes of new cluster created in step 3 above (say Current), run the same process as in step 2. If the found node is the leaf of parent's cluster a mesh link will be created between them. If the found node is a new node,

then a new cluster is created and the found node is nominated as new cluster header. Current will become the gateway for this newly created cluster. The Fig. 3(f) shows final four logical clusters with their cluster header as nodes 0, 5, 7 and 9.

5. Steps 3 and 4 will be repeated until no new neighbors are found by any cluster leaf node.

Fig. 3. Clustering of ZigBee Network

B. Minimum Battery Cost Routing in ME-AODV Accumulating path cost is used in ZigBee as the standard to

choose the routing. If length of path P is L then its accumulated path cost is:

{ }∑=

=+=

Li

iii DDCPC

11,)(

Where, C {Di, Di+1} implies that the link cost in path P from Di to Di+1 for link l can be calculated by:

⎪⎩

⎪⎨⎧

⎟⎟⎠

⎞⎜⎜⎝

⎛⎟⎟⎠

⎞⎜⎜⎝

⎛=4

1,7min

7

l

roundNρ

According to ZigBee specification ρl may be calculated by LQI (Link Quality Indication), however method for calculating ρl is not mentioned. Although above cost metric can reduce the total power consumption of the overall network, it does not reflect the lifetime of each node. This method of routing always selects least power cost routes which is a major disadvantage of this method. Nodes along these least-power cost routes tend to "die" soon because of the battery energy exhaustion. This is doubly harmful since the nodes that die early are precisely the ones that are needed most to maintain the network connectivity (and hence useful service life). Therefore, it would be better to use a higher power cost route if this routing solution avoids using nodes that have a small amount of remaining battery energy. To consume the energy in a more balance manner, an intuitive technique is to utilize cost function based on the node's remaining battery capacity. In Minimum Battery Cost Routing (MBCR) [17] the battery cost C for a particular route π with n nodes is defined as:

∑=

=

=ni

i iCC

1

1π

Ci is the remaining battery capacity of node. As the battery capacity decreases, the value of cost function for node n will increase. Therefore to find a route with maximum remaining battery capacity we should select a route π that has minimum battery cost C π = min{Ci | i Є A} where A is the set containing all possible routes. Since battery capacity is directly incorporated into the routing protocol, this metric prevents hosts from being overused, thereby increasing their lifetime and the time until the network is partitioned.

In order to incorporate MBCR into AODV, we have modified existing AODV protocol used in ZigBee. We have used the remaining battery power of the node as the cost function instead of the hop count used in traditional AODV. Whenever a router forwards route request or route reply packet it adds the reciprocal of its remaining battery power to the received packet and forwards it.

C. Multipath Routing in ME-AODV Multipath routing has been explored in several different

contexts. Traditional circuit switched telephone networks use a type of Multipath routing called alternate path routing [15]. In alternate path routing, each source node and destination node have a set of paths (or multipath), which consist of a primary path and one or more alternate paths. Multiple paths can provide load balancing, fault-tolerance and higher aggregate bandwidth. Load balancing can be achieved by spreading the traffic along multiple routes which eventually increases the network lifetime. From a fault tolerance perspective, Multipath routing can provide route resilience.

Ad hoc On-demand Multipath Distance Vector Routing (AOMDV) is an extension to the AODV protocol for computing multiple loop-free and link-disjoint paths. We have used a similar technique what is employed in AOMDV to construct the multiple path entries for a particular destination. To keep track of multiple routes, the routing entries for each destination contain a list of the next-hops along with the corresponding route cost (summation of reciprocal of remaining battery power of all the nodes participating in that path). Each duplicate route advertisement received by a node defines an alternate path to the destination. A router maintains route entry for at most MAX-PER-DEST-ENTRY (3, in our case) paths at a time. A router node i only accepts new advertised route φ if it has lower cost than all existing alternate paths in the route table. Let be the cost of received route φ

then it will be added to route list for the destination d if:

dCφ

( )di

di

di

d listrouteCCC _|min ∈<φ

Where, route_listid is the list of alternate route path for the

destination d at router node i. The router uses all available alternate routes in a round robin fashion in order to evenly distribute the energy consumption of the node over the entire network.

D. ME-AODV ROUTE DISCOVERY When a node wants to find route to another node, it initiates

a route discovery and sends a route discovery packet according to the following rules:

1. If source cluster Id is equal to destination cluster Id with a mesh link between them, no route request will be initiated-the data will be sent directly to the destination. In case a mesh link does not exist a unicast route request packet will be sent to the cluster head. If source cluster id is not equal to the destination cluster id the same procedure is followed as per traditional AODV with the following modification. The Time To Live (TTL) value is set equal to last known hop count from the expired route entry. This technique prevents unnecessary network-wide dissemination of route requests.

2. When a route request is received the following situations may occur: The device is the end device. In this case the route request packet is ignored.

a. The device is not an end device but a member of the destination cluster. This implies that the route request has reached its destination cluster and further propagation of route request may be blocked. This is achieved by reducing TTL value to 2 and then broadcast the route request.

b. The device is cluster header but not of the destination cluster. In this case device will simply decrement the TTL value and add the reciprocal of its remaining battery power to

the forward cost and broadcast it. c. The device is gateway for the destination

cluster. In this case, the packet will be forwarded as per traditional AODV.

d. The device is gateway for a cluster other than the destination cluster. The route request packet is ignored.

3. The route reply generation and forwarding is same as the AODV routing, except that the reciprocal of remaining battery power is used as the cost function instead of hop count.

4. On receipt of route reply at the source node, the source node updates its routing table if battery cost function of newly discovered route is better than all existing alternate routes. The source node also notifies the cluster head (if it is not the cluster head itself) about this newly discovered route information.

5. The cluster head broadcasts route information received in step 4 to all cluster members using cluster level broadcast (TTL = 1) so that all members of the cluster can share the route information to the destination.

IV. PERFORMANCE EVALUATION

A. Simulation Model We simulated a ZigBee network in ns-2 over IEEE 802.15.4

MAC/PHY module. In order to evaluate the performance of our ME-AODV routing we used a 50 X 50 m2 simulation area with 10 meter transmission range. The simulation run were carried out with different values of CM and RM. The results obtained from the simulation run are presented with CM, RM as 7 and 4 respectively. This was considered to be most "general" ZigBee network topology for our simulation scenario. The packet size was kept at 80 bytes. We used constant bit rate (CBR) as traffic source with average packet rate 0.5 packet/sec. The receiver and transmitter power were kept at 0.3 mW. The simulations were run for 100 sec for different source-destination traffic pair and various node positioning.

B. Performance Result In this section, we evaluate the efficiency of our routing

protocol. In earlier work [12] a ns-2 simulation model for IEEE 802.15.4 has been developed. Since IEEE 802.15.4 based AODV in ns-2 is similar to the network model that the ZigBee specification describes, we compared the performance of our ME-AODV protocol with it. We have taken three performance metrics (i) routing overhead, (ii) remaining node energy of all nodes after simulation and (iii) packet delivery ratio to evaluate the performance of our routing protocol.

1. Node Energy: The remaining node energy of all routers at the end of simulation has been plotted in Fig. 4. The graph shows that ME-AODV has distributed overall energy over

the entire network in a more balanced way. For example, due to multiple path routing, excessive energy consumption of node 4 has been shared by node 3. Similarly between node 12 shares the energy consumption of node 9.

2. Routing Overhead: The total number of packets used for the routing process is taken as the routing overhead in our study. The routing overhead shown in Fig. 5 shows the characteristics of ME-AODV and AODV with regard to routing overhead. On the average ME-AODV reduces the routing overhead by 30 percent as compared to AODV. This is because of route request pruning at the cluster level and less route discovery.

3. Packet Delivery Ratio: The ratio of data packets delivered to the destination and the data packets generated by the CBR sources are taken packet delivery ratio in our study. As shown in Fig. 6, ME-AODV outperforms the AODV because of limited congestion due to less routing overhead.

Fig. 4. Remaining energy of nodes after simulation

Fig. 5. Routing overhead for varying number of nodes

V. CONCLUSION In this paper, we have proposed a Multipath Energy Aware

AODV (ME-AODV) routing to improve the performance of existing ZigBee routing protocol. ME-AODV divides the ZigBee network into logical clusters. The proposed algorithm

exploits this logical cluster information to reduce the routing overhead. Along with the clustering technique a blend of Multipath routing and Minimum Battery Cost Routing (MBCR) has also been incorporated to increase the lifetime of network by load balancing of energy consumption. As a result ME-AODV makes an effort to reduce the number of route discovery and contribute on improving the network performances such as Network Lifetime, Packet Delivery Ratio and Routing Overhead. Simulation shows that ME-AODV significantly improves above mentioned network performance parameters.

Fig. 6. Packet Delivery Ratio for varying number of nodes

REFERENCES [1] IEEE Computer Society, "Part 5.4: Wireless Medium Access Control and

Physical Layer Specifications for Low Rate Wireless Personal Area Networks (LR-WPAN)", 1 Oct. 2003.

[2] P. Kinney, "ZigBee Technology: Wireless Control that Simple Works", Communication Design Conference" Oct. 2003

[3] ZigBee Alliance, "ZigBee Specification: ZigBee Document 053474r06 Version 1.0", 14 Dec 2004

[4] C. E. Perkins and E. M. Belding-Royer and S. R. Das, "Ad-hoc On-Demand Distance Vector Routing", Internet-Draft, IETF, March. 2002

[5] K. Jain, J. Padhye, V. Padamanabhan and L. Qiu, "Impact of Interference on Multi-Hop Wireless Network Performance", ACM Annual International Conference on Mobile Computing and Networking (MOBICOM), Sep, 2003, pp. 66-80

[6] S. Lee, E.M. Belding-Royer and C.E. Perkins, "Scalability study of ad hoc on-demand distance vector routing protocol", International Journal on Network Management (IJNM), to appear 2003

[7] C. K. Toh, "Maximum Battery Life Routing", IEEE Communication Magazine, June, 2001

[8] C. K. Toh, "Maximum Battery Life Routing", IEEE Communication Magazine, June, 2001

[9] M.K. Marina and S.R Das, "On-Demand Multipath Distance Vector Routing in Ad Hoc Networks", In Proceedings of IEEE International Conference on Network Protocols (ICNP), Pages 14-23, 2001

[10] Ian D. Chakeres and Elizabeth M. Belding-Royer. "AODV Routing Protocol Implementation Design." Proceedings of the International Workshop on Wireless Ad Hoc Networking (WWAN), Tokyo, Japan, March, 2004

[11] The Network Simulator - ns-2, http://www.isi.edu/nsnam/ns [12] Jianliang Zheng and Myung J. Lee, "A Comprehensive Performance

Study of IEEE 802.15.4", IEEE 802.15.4 draft} [13] Jose A. Gutierrez, Vinay Mitter, "IEEE 802.15.4: A Developing standard

for Low-Power Low-Cost Wireless Personal Area Network",IEEE Network, September, 2001.

[14] Ian D. Chakeres and Elizabeth M. Belding-Royer. "AODV Routing Protocol Implementation Design", Proceedings of the International

Workshop on Wireless Ad Hoc Networking (WWAN), Tokyo, Japan, March 2004

[15] Marc R. Pearlman, Zygmunt J. Haas, Peter Sholander and Siamak S. Tabrizi, "On Impact of Alternate Path Routing for Load Balancing in Mobile Ad Hoc Networks", In Proceedings of the ACM MobiHoc, pages 3-10, 2000