2011 Fifth International Conference on Innovative Mobile and Internet Services in Ubiquitous Computing
A Role-oriented Routing Algorithm Applied for Wireless Heterogeneous Networks Hsing-Chung Chen
Kanoksak Wattanachote
Department of Computer Science and Information Engineering, Asia University Taichung, Taiwan
[email protected] also
[email protected]
Institute of Computer Science and Information Engineering, Asia University Taichung, Taiwan
[email protected],
[email protected] information, it will calculate its shortest path from each source node to destination node. As shown in the Figure 1 (b), the different shortest route result is 1 → 4 → 3. In this paper, we propose a role-oriented routing protocol based on Knowledge Discovery (KD) agent with maintaining n-bits map routing table (n-bits MRT), which is suitable for WHNs environment. It is a role-oriented routing protocol which can get more flexible and achieve fault tolerant for packet routing in WHNs. In addition, the proposed routing protocol is suitable to the environment of Digital Home Wireless Heterogeneous Networks (DHWHNs). The rest of the paper is organized as below. In Sec. II, we introduce the related works. Our scheme is proposed in Sec. III. In Sec. IV, discuss and analyze for fault tolerant for packet routing and control packet overheads. Finally, our conclusions are presented in Sec. V.
Abstract—In recent years, on the researches in various routing protocols, applied to wireless heterogeneous networks, become more and more important. In this paper, we propose a roleoriented routing protocol applied for wireless heterogeneous networks based on a selected knowledge discovery agent with maintaining an n-bits map table. The routing protocol can be applied to digital home wireless heterogeneous networks. It is a role-oriented routing protocol which can get more flexible and achieve fault tolerant for packet routing in wireless heterogeneous networks. Keywords-wireless heterogeneous networks; role-oriented; knowledge discovery agent; digital home networks
I.
INTRODUCTION
Due to Internet has been development rapidly, various routing protocols, in wireless heterogeneous networks (WHNs), were proposed by researchers [4]. There are various routing protocols for routing packet data in Internet. The routing protocols can be categorized into two types; one is the Global Routing (GR) algorithm, and the other one is the Decentralized Routing (DR) algorithm [6].
(a)
II.
REVIEW OF KDATB ROUTING PROTOCOL
In 2008, Lee et al. [7] proposed a KDATB routing protocol, in which selected agent node will gather all network topology information, and then distribute a constructing bit-map routing table (BRT) to all other nodes. In BRT, a connecting state between two neighboring nodes is expressed as a binary bit. Through with the BRT, each node can decrease the overheads for maintaining its routing table [7].
(b)
Figure 1. A network topology demonstration, a source node is node 1 and a destination node is node 3. (a) The shortest path discovered by global routing algorithm (GR). (b) The shortest path discovered by decentralized routing algorithm (DR).
By way of GR algorithm as shown in Figure 1 (a), the routing path of minimum cost will be calculated between source node and destination node by gathering the topology information from global network, such as the knowledge discovery agent for topology bit-map (KDATB) algorithm [7]. The shortest route for the network topology in Figure 1 (a) is 1 → 2 → 3, if and only if node 1 plays a role as a source node and node 3 is a destination node. Whilst, both Routing Information Protocol (RIP) algorithm [3] and Ad hoc On-Demand Distance Vector Routing (AODV) algorithm [1, 2] are typical DR algorithms, in which each node only knows the costs which directly connect to other neighbor nodes. And when all nodes exchange the topology
978-0-7695-4372-7/11 $26.00 © 2011 IEEE DOI 10.1109/IMIS.2011.145
Figure 2. A network topology presents an agent node is receiving ‘Hello message’ from neighbor nodes in network. (Note: node 1 expresses as an agent node.)
For example, a network topology is shown as Figure 2, we suppose the node 1 is an agent node which directly
476 420
create the role-oriented routing protocol with n-bits MRT, we suppose there are n pre-defined roles, which are expressed as a role set Ra, b = { r1, r2, ..., ri , ..., rn−1, rn }, where ri is the one role which assigns to a node, and i is the index number of the i-th role, i ∈ {1, 2, ..., n − 1, n } and ri ∈ Ra, b. Thus, the n pre-defined roles Ra, b are corresponding to n binary bits Sa, b = { s1, s2, ..., si , ..., sn−1, sn } in n-bits MRT. When a role ri is assigned to a node, i.e. a smart phone, if a node is as the intermediate node for routing the role-driven event packets, its corresponding bit will be set as si = 1 for both nodes a or b, i.e. Sa, b = { s1, s2, ..., si = 1, ..., sn−1, sn } in its n-bits MRT. Otherwise, its corresponding bit will be set as si = 0 for node a or b, i.e. Sa, b = { s1, s2, ..., si = 0, ..., sn−1, sn } in its n-bits MRT. For example, we suppose Ra, b is a role set, Ra, b = { r1, r2, r3, r4, r5, r6, r7, r8 } ⇔ Sa, b = {0, 1, 1, 0, 0, 0, 0, 0}, can be expressed as {01100000} which these eight digital bits are corresponding to eight pre-defined roles. When a node states the two connections depended on it acting as the two roles ri, rj individually, the n-bits MRT will be marked both two routes via setting the directly connection states between itself and the next-hop neighboring node for the two roles, individually.
connects node 2 and node 3. The other nodes are node 4, 5, 6, 7 and 8. In Figure 2, all nodes will send a periodic ‘Hello message’ [5, 7] in order to confirm the existence of neighboring nodes. After receiving a ‘Hello message’ from neighboring nodes, the agent node will add into its neighbor information table [7] which consist the gathered information from all neighboring nodes. To construct the network topology, the agent node will broadcast a topology query message to all neighboring nodes. By this way, it creates a self-centered topology tree [7]. The node, which has received the topology query message, will assign the node which has sent the message to it as the parent-node, and all other neighboring nodes are assigned to be its child-nodes. After receiving a topology query message [7], the node will gather all information from sub-tree or said child-nodes and send a topology reply message [7] to its parent-node. The topology query message consists of the node’s ID, battery life-time and connectivity information [7]. When receiving topology reply messages from other nodes, the KD agent node will collect the topology information for all nodes. TABLE I.
THE CORRESPONDING BRT FOR NETWORK TOPOLOGY SHOWN ON FIGURE 2.
Node's IP Address The i-th role identity Node's Sequence Number Life time Figure 3. Message format.
In our protocol, one border router acts as an elected KD agent for each i-th role, which is used to gather information and deliver the knowledge data formed into a network topology as the form of an n-bits MRT to all nodes. The main works of the KD agent are to gather, calculate and distribute n-roles network topology knowledge. All nodes periodically broadcast a ‘Hello message’ in order to discover neighboring nodes for each role. The format of this ‘Hello message’ is shown in Figure 3.
After collecting all topology information, the KD agent node will create a BRT, for example, Table 1 shows a bitmap that expresses the network topology shown on Figure 2, where “1” is expressed as “one-hop distance connection” and “0” is expressed as “non one-hop distance connection”. Finally, the KD agent node uses the topology advertisement message [7] in which the bit-map table is carried and sent to other nodes. When a node needs to deliver a packet to other node, it will check the routing path by using the bit-map table. It can quickly establish multiple routing paths which include the shortest path from source node to destination node. The bit-map table represents a graph which is shown as the connected link state for all nodes in the network. III.
Figure 4. The i-th role neighbor information table for node n1.
OUR SCHEME
Figure 5. The i-th role topology query message format.
In this paper, we first propose a role-oriented routing protocol for WHNs environment. There are m exiting nodes in WHNs. A node a or b is one of m nodes and a ≠ b. Within the n-bits MRT, a connecting state si corresponding to a role ri between two neighboring nodes a or b, which is expressed as a binary bit si ={0, 1}, where “1” is expressed as “one-hop distance connection for the role ri” and “0” is expressed as “non one-hop distance connection for the role ri”. In order to
Figure 6. The i-th role topology reply message format.
421 477
TABLE II.
After receiving a ‘Hello message’ from neighboring nodes for the role r1, a node generates the neighbor information and gathers them into n-roles neighbor information table. For example, an n-roles neighbor information table is shown in Figure 4, which shows the table collecting by an intermediate node n1 based on Figure 7. In Figure 7, node b1 is the KD agent and node n1 has the following steps for gathering neighbor information: A. Receive topology query message from node b1. B. Assigns node b1 as the parent-node, and nodes l2 and n4 as child-nodes. C. Broadcasts a topology query message to child-nodes l2 and n4.
THE N-BITS MRT FOR ROLE-ORIENTED ROUTING PROTOCOL.
Note: Rx, y = { r1, r2, …, rn } = { s1s2…sn }.
The elected KD agent will send the n-bits MRT information to all nodes in the heterogeneous network via using a topology advertisement message. Figure 8 shows the topology advertisement message format for each i-th role. After each node has received the topology advertisement messages, it will check whether there are any differences between this and previous topology information in order to update the n-bits MRT. Figure 7. A topology of wireless heterogeneous network. (Note: b1~bp: border routers (knowledge discovery agents) ; n1~nq: the 1st network nodes; l1~lk: the 2nd network nodes; r1~rn: role identities)
Then, the KD agent node b1 broadcasts a query message to all neighboring nodes to collect information for constructing network topology of the i-th role, i.e. r1. The node transmitting the topology query message becomes the parent-node, and the receivers become the child-nodes for constructing network topology of the i-th role. Figure 5 shows the format of the query message. By this means it constructs the i-th role self-centered topology tree. The node which has received the topology query message assigns the node which has sent the message to be the parent-node, and all other neighboring nodes are assigned to be child-nodes. After receiving all topology reply messages from its child-nodes, a parent node records all information in the topology information table and transmits the information to its parent-node for the i-th role. Figure 6 shows the format of the i-th role topology reply message. After receiving topology information, the agent node constructs the n-bits MRT, shown in Table 2, for roleoriented routing protocol. Based on the n-bits MRT, the elected KD agent can be changed according to the following rules: A. The intermediate node is the preferred agent node. B. The agent node should have the longest battery lifetime. C. The agent node should have as many neighboring nodes as possible.
Figure 8. Advertisement message format.
When a node needs to transmit data to another node, it checks a routing path to the destination node by using the nbits MRT. It can quickly establish the connection. The n-bits MRT represents a graph which shows the network topology. The pseudo algorithm for the path search process is illustrated below. The path search algorithm for the i-th role-oriented routing protocol: Step 1: Perform “AND” operation with source node and destination node for the i-th role. 1a. if there is a path for the i-th role to the destination node, then go to step 3 1b. if there is no path for the i-th role, then go to Step 2. Step 2: Find neighboring node for the i-th role. 2a. Select an unchecked neighboring node for the i-th role, perform “AND” operation with destination. 422 478
(1) find a path for the i-th role to destination, and then go to Step 3. (2) if there is no path for the i-th role, go to Step 2a. 2b. when all neighbors assigning for the i-th role are checked, then select a neighbor and perform Step 2, recursively.
V. CONCLUSIONS In this paper, we first propose an n-bits map role-oriented routing protocol for WHNs urban safety environment. It is a role-driven routing protocol which can reduce the number of control packets and achieve fault tolerant for packet routing in WHNs.
Step 3: Use searched nodes assigning for the i-th role for the path, and finish. IV.
ACKNOWLEDGMENT This work was supported by National Science Council, Taiwan, Republic of China, under Grant NSC99-2221-E468-011.
DISCUSSIONS AND ANALYZE
In this subsection, our proposed protocol is compared with both of traditional AODV and R-AODV (Reverse RAODV), see (Kim et al., 2006), in the number of control packets, shown in Table 3. We define the relevant parameters as below: A. t : number of nodes in WHNs relied on route reply message. B. p+q+k : number of nodes in WHNs. C. n : number of roles in WHNs. TABLE III.
REFERENCES [1]
[2] [3] [4]
THE NUMBERS OF CONTROL PACKET.
[5]
[6]
We compare the number of control packets among AODV, R-AODV with our protocol as shown in Table 3. AODV needs to transmit the (p+q+k)-1+t control packets. And R-AODV needs to transmit the 2(p+q+k)-2 control packets. Because of dividing the WHNs into n distinct role groups, our protocol only needs ((p+q+k)/n)-1+t control packets. Therefore, our propose protocol can reduce the number of control packets. In other words, the proposed protocol can save the power of nodes in WHNs. If one path has interrupted, the intermediate node can easy choose a new path by using the path search algorithm for the i-th role-oriented routing protocol algorithm via the n-bits MRT. Therefore, our propose protocol can achieve fault tolerant mechanism.
[7]
423 479
Kim C., Talipov E. and Ahn B., A reverse AODV routing protocol in ad hoc mobile networks, in Proceedings of Embedded and Ubiquitous Computing, vol. 4097, 522-531, 2006. Perkins C. and Belding-Royer E., Ad hoc On-Demand Distance Vector (AODV) Routing, RFC 3561, 2003. Malkin G., RIP Version 2, RFC 2453. , 1998 Yasser G. and Hadeel E. K., A WSN/MANET hybrid protocol for routing data in heterogeneous wireless sensor networks, in Proceedings of Wireless Communications and Mobile Computing, 523 - 528. , 2008 Chakeres I. D. and Belding-Royer E. M., The utility of hello messages for determining link connectivity, International Symposium on Wireless Personal Multimedia Communications, vol. 2, 504-508, 2002. Kurose J. K. and Rose K. W., Computer Networking: A Top-Down Approach, 4th ed., New York: PEARSON, 2007. Lee S. S., Kim H. K. and Kim C. G., A knowledge discovery agent for a topology bit-map in ad hoc mobile networks, Journal of Universal Computer Science, vol. 14, no. 7, 1105-1117, 2008.
