Cross-Layer Power-Control-Based Real-Time Routing Protocol for Wireless Sensor Networks

Abstract

In order to solve contradictions between real-time applications of wireless sensor networks and limited energy of nodes, Cross-Layer Power Control Algorithm (CLPCA) is proposed in this paper. Moreover, Power-Control-Based Real-Time Routing Protocol (PCBRRP) is designed capable with CLPCA. In this protocol, the transmission power of the node is adjusted dynamically to increase the energy efficiency. The next hop node is selected via the link quality of the communication and the residual energy of every node. Simulation results show that the energy consumption of the network is reduced while real-time end-to-end data transmission is guaranteed.

1. Introduction

Wireless sensor networks (WSNs) are widely used in many areas, such as medical surveillance [1], security monitoring [2–4], and fire rescuing [5, 6]. All these applications impose real-time overheads on WSNs; that is, the data transmission tasks must be performed in limited time period. However, the energy stored in sensor nodes in WSNs is relatively limited, which is constrained by the deployment environment, and is difficult to be recharged [7, 8]. Therefore, the key issue of real-time applications is to use the limited amount of energy efficiently and to guarantee the real-time performance to a certain extent [9].

In WSNs, efficient routing protocols can be used to reduce unnecessary energy consumption in data transmission process. Researchers have proposed many routing strategies in order to meet real-time requirement. SPEED [10, 11] is a classic real-time routing protocol based on data transmission delay, and neighboring feedback strategy is adopted to ensure that network transmission rate is above global transmission rate threshold. This protocol achieves the goal of real-time data transmission, but it only defines a global transmission threshold; consequently, it cannot be suitable for different transmission delay requirements of different data packets. MMSPEED [12], which is an extended version of SPEED, divides the candidate nodes into multiple subsets according to the local transmission rate between current node and each candidate routing node. Each subset is a virtual subnet. As a result, a physical WSN can be logically divided into multiple subnets with different transmission rates. According to different transmission rate requirements of each data packet, MMSPEED allocates each data packet to different logical subnets. To some extent, this protocol can meet the requirements of different data packets with different transmission rates. However, both of them use the fixed transmission power and do not consider the residual energy of the node, which will yield the premature death of some nodes and, finally, harm the network lifetime.

2. Problem Description

In Figure 1, a WSN is modeled as an undirected connected graph, denoted as $G (V, L)$ , where Vrepresents the set of nodes and L represents the set of edges. The edge existing between node i and node j, denoted as $l (i, j) \in L$ , represents a wireless connection between node i and node j.

Figure 1

Undirected connected graph model of WSNs.

Definition 1.

When a packet is transmitted from the source node S to the destination node D, the default transmission rate of this packet is defined as follows:

\begin{matrix} v_{req} (S, D) = \frac{d (S, D)}{t_{req}} . \end{matrix}

(1)

When a packet is transmitted from the source node S to the candidate node i by power $p_{S}$ , the actual transmission rate of this packet is defined as follows:

\begin{matrix} v_{S \to i} (p_{S}) = \frac{d (S, D) - d (i, D)}{Dela y_{S \to i} (p_{S})}, \\ i \in F (i) ≜ {i ∣ d (S, D) - d (i, D) > 0, i \in N (i)} . \end{matrix}

(2)

Here, $d (S, D)$ is the Euclidean distance between source node S and destination node D; $d (i, D)$ is the Euclidean distance between node i and destination node D; $t_{req}$ is the preset transmission time of the packet; $Dela y_{S \to i} ({p_{S}}_{})$ is the transmission delay of source node S when node S sends data to node i with power ${p_{S}}_{}$ ; $F (i)$ is the candidate forwarding node sets of next jump recognized by node S; and $N (i)$ is the one-jump neighboring node sets of node S.

Definition 2.

When a packet is transmitted from source node S to destination node D, $E (S, D)$ represents the energy consumed by this transmission process, defined as follows:

\begin{array}{l} E (S, D) \approx E^{S \to i} (p_{S}) \times C^{S \to i} (p_{S}) \times \frac{d (S, D)}{d (S, D) - d (i, D)}, \\ i \in F (i) . \end{array}

(3)

Here, $E^{S \to i} ({p_{S}}_{})$ is the energy consumed in this transmission process when source node S sends a single packet to node i with power ${p_{S}}_{}$ ; $C^{S \to i} ({p_{S}}_{})$ is the number of packets in the transmission process when the source node S sends packets to node i with power ${p_{S}}_{}$ .

The main purpose of a real-time routing protocol is to ensure that data packets can be transmitted to the destination node within preset transmission time, while making energy consumption as small as possible. To achieve these goals, we first define some rules as follows.

Rule 1.

When a packet is transmitted from the source node S to the destination node D, during each jump, the packet should meet the condition as follows

\begin{matrix} v_{req} (S, D) \leq v_{S \to i} (p_{S}), \end{matrix}

(4)

where

v_{S \to i} (p_{S})

is the path which satisfies real-time requirements.

Here, $v_{req} (S, D)$ is the preset transmission rate of a packet when it is transmitted from the source node S to the destination node D; $v_{S \to i} (p_{S})$ is the actual transmission rate of a packet when it is transmitted from the source node S to the next hop node i by power $p_{S}$ . Under this rule, it can be guaranteed that each actual transmission rate of a packet through each hop is not less than the global preset transmission rate. In other words, this rule will ensure that a packet can be transmitted to the target node within the preset time $t_{req}$ , so as to guarantee the real-time data transmission requirements.

Rule 2.

Minimizing the $E (S, D)$ with certain routing strategy in consistence with Rule 1.

We assume that the number of hops from the source node to the destination node is n, and Rule 1 has ensured that the maximum transmission delay time of a single-hop is T. With these two provisions, we can theoretically guarantee that the upper bound of multihop transmission delay time from the source node to the destination node is $n T$ . Some existing routing protocols only follow Rule 1 and assume that transmission power of a sensor node is always fixed. However, through the theoretical analysis, Gomez and Campbell [13] have proved that dynamic adjustment of transmission power is better to improve working ability and energy efficiency compared with the fixed transmission power manner.

Aiming at overcoming the weaknesses of existing protocols, a Cross-Layer Power Control Algorithm (CLPCA) based on the two above rules and theories of Gomez et al. is proposed in this paper. Moreover, we also design the Power-Control-Based Real-Time Routing Protocol (PCBRRP) capable with CLPCA; that is, considering residual energy of forwarding node and the link quality of communication with real-time requirements, PCBRRP adopts real-time routing option strategy and effective neighboring nodes management mechanism. With these designs, PCBRRP protocol can ensure real-time transmission of end-to-end data successfully and reduce the network energy consumption.

3. Cross-Layer Power Control Algorithm (CLPCA)

Related studies have shown that energy consumption is extremely low in terms of perception module or processing module in a sensor node. Most energy is consumed by the communication module [14]. Power control technology in WSNs can significantly improve the energy consumption and real-time performance in network systems. Therefore, combining the background of real-time application and the status of every network layer, we can effectively use the power control based routing protocol to alleviate the trade-off between the real-time and energy consumption. Based on the number of optimal neighboring nodes, the Cross-Layer Power Control Algorithm (CLPCA) is proposed in this paper. Under the circumstance of ensuring network connectivity, this algorithm dynamically adjusts transmission power of one node and selects the minimum required emission power to communicate with its neighboring nodes to improve network energy efficiency. This algorithm also takes into account the multichannels power control in link layer and the power control strategy which is based on the number of optimal neighboring nodes in network layer. This algorithm is described in detail as follows.

Step 1.

Once initialized, node i will divide the original channels into signaling channel, forwarding data channel, and reversing ACK channel.

Step 2.

Node i in the signaling channel broadcasts a message MSG containing its own ID and sets the value of broadcasting cycle counter at the same time.

Step 3.

After receiving the MSG of node i, node j judges whether the value of signal strength at node i is higher than the value of its own signal strength. If yes, the minimum required emission power of node i will be calculated by the Friis [14] formula. This value along with node i will be written into the routing table, and node j will also send an ACKMSG containing its own ID to the node i through the signaling channel.

Step 4.

Node i checks the number of received ACKMSG, denoted as N, and sets this value as the number of its neighbors.

Step 5.

According to the relationship between N and the range of its neighboring nodes threshold, node i uses the power adjustment mechanism in LMA [15] to dynamically adjust its emission power.

Step 6.

From the candidate neighboring nodes, node i will carefully select node k that meets the forwarding requirements as its next hop node and uses the forwarding data channel to send data. If there are no forwarding nodes meeting the requirements, the procedure will go back to Step 5.

Step 7.

After receiving the data from node i, node k will send the data acknowledgment frame (ACK) to the reversing channel at the maximum emission power to fix the possible occurring unidirectional links.

Step 8.

If node i does not receive the data acknowledgment frame (ACK) from node k, then node i will go back to Step 2; otherwise, it will terminate the operation.

4. Power-Control-Based Real-Time Routing Protocol (PCBRRP)

By calculating the forwarding adaptation index of a node, denoted as $θ_{S - i}$ , PCBRRP protocol can make data packets select their optimal next hop node that satisfies the condition (4) to avoid the serious phenomenon of energy consumption in a local communication link and prolong network lifetime. If nodes in $F (i)$ do not meet the condition (4), PCBRRP protocol, combining with CLPCA, will adopt efficient neighboring nodes management mechanism to make reasonable decisions for forwarding or discarding the packets. PCBRRP protocol includes real-time routing options and neighboring nodes management.

4.1. Real-Time Routing Option

4.1.1. Delay Estimation

In order to calculate the actual transmission rate, PCBRRP protocol needs to know the transmission delay of adjacent nodes. Due to the constraints of node energy and bandwidth, using a specialized probe is not suitable for measuring single-hop data transmission delay. Therefore, in this paper, instead, we estimate the actual packet transmission delay. When source node S sends packets to the next hop node i by power $p_{S}$ , the transmission delay in this process is calculated as follows:

\begin{matrix} Dela y_{S \to i} (p_{S}) = T_{c} + T_{t} + T_{p} + T_{q} + T_{s} = \frac{T_{2} - T_{1}}{2} . \end{matrix}

(5)

Here, $T_{c}$ is the time for packets competing for channel occupation; $T_{t}$ is packet transmission time; $T_{p}$ is packet processing time; $T_{q}$ is the time delay in the transmitting queue; $T_{s}$ is node sleeping time. Node S will first record the time instance ( $T_{1}$ ) when packet enters MAC layer and then records the time instance ( $T_{2}$ ) when node S receives the ACK packet. Half of the difference between these two moments $(T_{1}, T_{2})$ is the single-hop transmission delay time. In order to dynamically allocate forwarding rates to the packets, PCBRRP protocol uses the moving weighted average method to update $Dela y_{S \to i} (p_{S})$ , calculated as follows:

\begin{array}{l} Dela y_{S \to i}^{t + 1} (p_{S}) \\ = α Dela y_{S \to i}^{t} (p_{S}) \\ + \frac{1 - α}{T} \sum_{k = \max (1, t - T)}^{t - 1} Dela y_{S \to i}^{k} (p_{S}), 0 \leq α \leq 1, \end{array}

(6)

where T is the time window for competition, the value of parameter α is associated with delay change. α has a larger value when the network delay changes significantly; otherwise, it has a smaller value.

4.1.2. Link Quality

Link quality will obviously affect the ratio of successful transmissions directly. When real-time routing option strategy selects the next hop nodes in BPCRRP protocol, it will take into account the communication link quality to ensure real-time reliable transmission of data packets. Moreover, BPCRRP protocol also adopts packet reception rate (PRR) to approximatively estimate the communication link quality. The formula of PRR in literature [16] is calculated as follows:

\begin{array}{l} PRR = [1 - (\frac{8}{15}) (\frac{1}{16}) \sum_{j = 2}^{16} {(- 1)}^{j} (\begin{pmatrix} 16 \\ j \end{pmatrix}) \\ {\times \exp (20 γ (d) (\frac{1}{j} - 1))]}^{176}, \\ γ (d) = P_{t} - PL (d) - S_{r} \\ PL (d) = PL (d_{0}) + 10 n \log (\frac{d}{d_{0}}) + X_{σ} . \end{array}

(7)

Here, $γ (d)$ is the signal to noise ratio; $PL (d)$ is the value of communication link loss; $P_{t}$ is the transmission power of source node; $S_{r}$ is the sensitivity value of destination node; d is the Euclidean distance between source node and destination node, $d_{0}$ , n; and $X_{σ}$ is the set of all the constants.

If the value of d is determined in the formula (7), the PRR will be updated only when the node transmission power changes. The update process will follows the formula listed below:

\begin{array}{l} PRR (t + 1) = α PRR (t) \\ + (1 - α) \frac{1}{t - 1} \sum_{k = 1}^{t - 1} PRR (k), 0 \leq α \leq 1, \end{array}

(8)

where α is a random number representing the current network traffic. Moreover, if the network traffic is heavier, α will be bigger. Otherwise, it will be smaller.

4.1.3. Routing Selection

When source node S has a packet to send, it will first check whether the sink node is the next-hop node. If so, node S will send packets directly to the sink node; otherwise, node S will calculate forwarding adaptation index $θ_{S - i}$ of all nodes in $F (i)$ that satisfies the conditions (4) and select the node with the maximum value ( $θ_{S - i}$ ) as its next hop node. The formula of $θ_{S - i}$ is shown as follows:

\begin{matrix} θ_{S \to i} = α \times \frac{v_{S \to i} (p_{S})}{\sum_{i \in F (i)}^{} v_{S \to i} (p_{S})} + β \times \frac{E_{i}}{E^{S \to i} (p_{S})} + γ \times PRR, \\ α + β + γ = 1 . \end{matrix}

(9)

Here, α, β, and γ are weighted factors.

v_{S \to i} (p_{S})

is the actual transmission rate of a packet when it is transmitted from the source node S to the next hop node i by power

p_{S}

E^{S \to i} (p_{S})

is the energy consumed in the process when source node S sends a single packet to node i with power

p_{S}

E_{i}

is the current residual energy of node i. Thus, it can be seen that the forwarding adaptation index is calculated by comprehensively considering transmission rate of a packet, residual energy of the next hop node, quality of communication link, and other factors.

If all the nodes in $F (i)$ do not meet the conditions (4), then node S will initiate the neighboring nodes management mechanism to decide if this packet is to be forwarded or discarded.

4.2. Neighboring Nodes Manager

4.2.1. Power Dynamic Adjustment

While ensuring data real-time transmission, energy consumption in communication link $(E (S, D))$ should be as small as possible. In order to achieve this goal, node S will take advantage of power dynamic adjustment mechanism of neighboring nodes to select its next-hop node.

(1) Increase the node transmission power: if less than 20% of nodes in $F (i)$ satisfy the condition (4), BPCRRP protocol will select some nodes from $F (i)$ to increase their original transmission power by α ( $α > 1$ ) times. These nodes need to meet the following conditions. (i)

Transmission power does not reach the maximum value.

(ii)

The nodes have a low workload.

(iii)

The change of $F (i)$ does not lead to the increase of subsequent transmission delay.

Then, source node S will select the maximum value ( $θ_{S - i}$ ) from these nodes as the next hop node. The value of α should be chosen reasonably since a large α will inevitably cause energy waste, while a small α may not be able to guarantee the real-time transmission of data packets.

(2) Reduce the node transmission power: if more than 80% of nodes in $F (i)$ satisfy the condition (4), BPCRRP protocol will linearly decrease their original transmission power by β times to reduce the energy consumption. These nodes need to meet the following conditions. (i)

Transmission power does not reach the minimum.

(ii)

The nodes have a low workload.

(iii)

The change $F (i)$ does not lead to the increase of transmission delay.

Then, source node S will select the maximum value ( $θ_{S - i}$ ) from these nodes as the next-hop node. The value of β should be chosen reasonably since a small β will inevitably cause energy waste, while a large β may not be able to guarantee the real-time transmission of data packets.

4.2.2. Neighboring Nodes Discovery

In $F (i)$ , if source node S still does not find the next hop node that meets the forwarding conditions by power dynamic adjustment mechanism of neighboring nodes, node S will use the neighboring discovery strategy to find the new forwarding next hop node in $N (i)$ . In detail, node S broadcasts a packet of Request to Route (RTR) by medium power P. In order to reduce the network energy consumption caused by the network competition, nodes satisfying the following conditions are qualified to reply. (1)

$d (S, D) - d (j, D) > 0, j \notin F (i)$ .

(2)

Node j should satisfy condition (4) as well as condition (1).

Therefore, RTR packet should piggyback source node S, position coordinates of destination node D number, and ID information of each node in $F (i)$ . When a node receives the packet RTR, it will first compare its own ID number with the ID number of each node in $F (i)$ within RTR packet fields. If the result is equal, the RTR package will be discarded. Otherwise, according to the position coordinates of the source node S and destination node D in RTR packet fields, we will calculate the distance and rate, then determine whether it meets the condition (1) or (2). If satisfied, then the node will reply the routing request; otherwise, the node will discard the RTR packet. Node S will also add the node which replies its routing request into the $F (i)$ and at the same time deletes the lower frequency of forwarding nodes in $F (i)$ .

5. Simulation Results

We use NS2 to conduct simulation experiments for PCBRRP, compared with both the SPEED and MMSPEED. The simulation scenarios are as follows. 121 nodes are uniformly distributed in the coordinates between (0 m, 0 m) and (100 m, 100 m). The coordinates of these 121 nodes are already known. Among these nodes, the coordinates of the sink node are (50 m, 50 m). The node transmission power range is (−20 dB, 10 dB), corresponding to the range of their power consumption of (3.5 mA, 21.5 mA). In addition, the initial energy of every node is 3.3 J, the value of N is taken as 6, data rate is 12 p/s, α, β, and γ in the formula (4) are 0.2, 0.2, and 0.6, respectively, and α and β in the power dynamic adjustment mechanism of neighboring nodes are 1.25 and 1.20, respectively.

PRR is used to approximate the quality of communication link in BPCRRP. If the transmission power of a node changes, the value of PRR will be updated by using formula (8). With this updated PRR, it can be ensured that if excessive congestion occurs because of heavy network traffic, the next hop node can be chosen scientifically in BPCRRP.

In order to inspect the real-time performance and energy efficiency of these protocols, we use packet loss rate and energy consumption of a single packet as the measuring indicators. Now, we show the simulation results in Figures 2 and 3.

Figure 2

Packet loss rate comparison chart.

Figure 3

Energy consumption comparison chart.

Figure 2(a) is the packet loss rate comparison chart with different transmission delays for different protocols. In the case of small transmission delay time, the packet loss rate of PCBRRP is decreased by almost 12.5% and 30%, compared with the other two protocols. When packet transmission delay reaches 250 s, the packet loss rates of these three protocols are all within 20%. As shown in Figure 2(b), the packet transmission delays are all controlled around 250 s for these protocols. After the network operates for 150 s, packet loss rate compared with different running time is also clearly known. Within 300 s, packet loss rate remains within 40% in PCBRRP, and packet loss rate is decreased by almost 11% and 29.5% with respect to data real-time transmission, compared with the other two protocols.

Figure 3(a) is a single packet node energy consumption comparison chart with different transmission delays for different protocols. Clearly, under the precondition of ensuring data transmission delay, PCBRRP can reduce the node energy consumption to the minimum value. As the running time is gradually extending, Figure 3(b) shows that growth curve of a single node energy consumption in the network is lower by using PCBRRP. Therefore, PCBRRP can reduce energy consumption of network nodes.

6. Conclusions

In order to alleviate the trade-off between real-time applications of wireless sensor networks and limited energy of nodes, the Cross-Layer Power Control Algorithm (CLPCA) and the Power-Control-Based Real-Time Routing Protocol (PCBRRP) are proposed in this paper. PCBRRP can effectively ensure the real-time data transmission according to the simulation result. With node power dynamic adjustment mechanism, we can also reduce the network transmission energy consumption to the minimum theoretically. However, the asymmetric emission power will also bear a range of adverse effects [17], such as serious phenomenon of hidden stations in a network, conflicts increase, and fairness decrease. Therefore, minimizing these adverse effects and balancing the network performance are the issues for further researches.

Footnotes

Acknowledgments

The authors acknowledge the financial support by the Special Fund Project of National IOT Development (no. MOIIT.(2012)583), Doctoral Fund of Ministry of Education of China (no. 20100111110004, no. 20120111110001), Natural Science Foundation of Jiangsu Province (no. BK2011236), Natural Science Foundation of Anhui Province (no. 1208085QF113), and International S&T Cooperation Program of Anhui Province of China (no. 1303063009).

References

Guangzhao

Dalong

Software design of WSN node in health care system

Institute of Information Engineering 2010 36 1 256 259

Vicaire

Yant

Luo

Zhou

Stoleru

Cao

Stankovic

J. A.

Abdelzaher

Achieving real-time target tracking using wireless sensor networks

Proceedings of the 12th IEEE Real-Time and Embedded Technology and Applications Symposium

April 2006

37 48

2-s2.0-33749595383

10.1109/RTAS.2006.9

Applications of environmental security monitoring based on WSN in substation

Proceedings of the 4th International Congress on Image and Signal Processing (CISP '11)

October 2011

2806 2808

2-s2.0-84855573243

10.1109/CISP.2011.6100752

Shengyu

Cong

Guang

Ding

W. U.

Indoor assist system for rescue activity based on wireless sensor

Journal of University of Electronic Science and Technology of China 2010 39 4 107 110

Yanning

Raj

Lili

May

Opportunities for WSN for facilitating fire emergency response

Proceedings of the 5th International Conference of Information and Automation for Sustainability (ICIAFs '10)

2010

81 86

Yingli

Lingqin

Tingting

Monitoring system for forest fire based on wireless sensor network

Proceedings of the 10th World Congress of Intelligent Control and Automation (WCICA '12)

2012

4245 4248

Yang

Kang

An energy-efficient and fault-tolerant convergecast protocol in wireless sensor networks

International Journal of Distributed Sensor Networks 2012 1 8

Yang

Sun

Taheri

Zomaya

A. Y.

DLS: a dynamic local stitching mechanism to rectify transmitting path fragments in wireless sensor networks

Journal of Network and Computer Applications 2013 36 1 306 315

Kulakowski

Limited resources in ambient systems for disaster scenarios

Proceedings of the 10th IEEE SENSORS Conference (SENSORS '11)

October 2011

655 658

2-s2.0-84856943768

10.1109/ICSENS.2011.6127021

10.

Felemban

Lee

C.-G.

Ekici

Boder

Vural

Probabilistic QoS guarantee in reliability and timeliness domains in wireless sensor networks

Proceedings of the 24th Annual Joint Conference of the IEEE Computer and Communications Societie

March 2005

4 2646 2657

2-s2.0-25644460065

11.

Kordafshari

M. S.

Pourkabirian

Faez

Rahimabadi

A. M.

Energy-efficient speed routing protocol for wireless sensor networks

Proceedings of the 5th Advanced International Conference on Telecommunications (AICT '09)

May 2009

267 271

2-s2.0-70349742227

10.1109/AICT.2009.52

12.

Darabi

Yazdani

Fatemi

Multimedia-aware MMSPEED: a routing solution for video transmission in WMSN

2008 2nd International Symposium on Advanced Networks and Telecommunication Systems, ANTS 2008

December 2008

2-s2.0-67650518018

10.1109/ANTS.2008.4937803

13.

Gomez

Campbell

A. T.

A case for variable-range transmission power control in wireless multihop networks

Proceedings of the 23rd AnnualJoint Conference of the IEEE Computer and Communications Societies (INFOCOM '04)

March 2004

1425 1436

14.

Valli

Dananjayan

Power control with QoS constraints in WSN using game theoretic approach

Proceedings of the 3rd International Conference of Computing Communication and Networking Technologies (ICCCNT '12)

2012

15.

Kirousis

L. M.

Kranakis

Krizanc

Pelc

Power consumption in packet radio networks

STACS 97 1997 1200 363 374 Lecture Notes in Computer Science

16.

IEEE 802. 15. 4 Standard Part 15. 4: wireless medium access control (MAC) and physical layer (PHY) specifications for Low-Rate Wireless Personal Area Networks (LR-WPANs), 2003

17.

Alayev

Chen Fangfei Yun

Throughput maximization in mobile WSN scheduling with power control and rate selection

Proceedings of the 8rd International Conference of Distributed Computing in Sensor Systems (DCOSS '12)

2012

33 40