Sage Journals: Discover world-class research

Abstract

Hybrid wireless mesh networks are suitable to construct emergency communication networks after disasters in underground mines. The routing decision in emergency scene is more difficult to give an accurate mathematical description due to the constraints of various data types, different data transmission requirements, and multi-parameters. Based on the fuzzy decision theory, this article has proposed a fuzzy-logic-based data-differentiated service supported routing protocol. Through the use of the adaptive fuzzy decision system, fuzzy-logic-based data-differentiated service supported routing protocol can provide data-differentiated services and make optimized routing decisions to satisfy the transmission requirements of different data types. In addition, a path soft handoff strategy has been proposed to maintain continuous data transmission when the path quality deteriorates. Based on NS2, we set three transmission scenarios (transmitting emergency data, regular data, or mixed data) to test the performances of fuzzy-logic-based data-differentiated service supported routing protocol, ad hoc on-demand distance vector, FUZZY-ad hoc on-demand distance vector, and multi-criteria routing metric. The results show that the fuzzy-logic-based data-differentiated service supported routing protocol has a higher delivery ratio and lower end-to-end delay when transmitting emergency data. When transmitting regular data, fuzzy-logic-based data-differentiated service supported routing protocol has achieved higher throughput and longer network lifetime than that of similar algorithms.

Keywords

Emergency communication network routing protocol fuzzy decision underground mine wireless mesh networks differentiated service

Introduction

Coal mining environment is complex and accompanied by many risk factors such as high dust content, loud noises, and toxic gases. Therefore, in order to predict accidents and disasters, it is necessary to construct a communication network to monitor the status of miners, equipment, and the environment.¹ The traditional network in an underground mine is composed of network equipment (communication cables, Ethernet switches, routers, etc.) and communication clients (sensor nodes, personnel location nodes, mobile intelligent clients, etc.). However, clients and equipment in the network may be damaged once a disaster occurs. As a result, the data transmission capacity of the network will be seriously degenerated and emergency rescue communications cannot be supported after a disaster. Therefore, it is necessary to build an emergency communication network in an underground mine (ECNUM) through the remaining network equipment and communication clients.

The construction scene of the ECNUM after tunnel obstruction is shown in Figure 1. The tunnel blockage cannot be quickly opened up as usual, and rescue workers are unable to enter the affected area in a short time. To obtain the status of the affected area, the commonly used method is to drill in the blocking area and install a wireless access point (AP) or gateway through the hole to connect the ground backbone network with the ECNUM.²

Figure 1.

Construction of the ECNUM.

Wireless mesh networks (WMNs), as a new type of wireless network, have broad application prospects. They can be classified into three types: backbone WMNs, terminal WMNs, and hybrid WMNs (HWMNs).³ HWMNs consist of mesh routers (MRs) and mesh clients (MCs). MCs can perform routing functionalities and provide end-user applications to customers. HWMNs in long and narrow tunnels mostly have long-chain structures.⁴ The network topology after a disaster is shown in Figure 2, where GW is a gateway node installed in the hole at the blocking area. HWMNs have strong self-organizing and self-repairing abilities. When MRs are partly or totally unavailable, the network can be reconstructed through MCs to continue the transmission of data as shown in Figure 2. Therefore, HWMNs can be used to build an ECNUM in a complex disaster environment because of their high robustness.⁵

Figure 2.

Topology of the ECNUM after a disaster.

However, there have been few efficient dedicated routing protocols for ECNUM. The existing routing protocols did not consider the characteristics of post-disaster underground mines. For example, there are two kinds of data (regular data and emergency data) transmitted in the ECNUM. The network quality of service (QoS) requirements for emergency data and regular data are different. Emergency data are small and have strong burstiness, which demands real-time and reliable transmissions. The data stream of regular data is large and continuous, which will consume more energy of the network equipment. Therefore, it is necessary to provide differentiated services for different types of data when designing the ECNUM routing protocol. In addition, disasters may damage the underground power supply system, and energy constraints will be more prominent. Therefore, when regular data are transmitted in the ECNUM, energy balance and load balance should be achieved to maximize the network lifetime through data routing.

QoS is usually used to evaluate a transmission path. The QoS of ECNUM involves many aspects, including delay, delivery ratio, throughput, and lifetime. However, the relationship of these parameters is often complex, and the description of path quality is fuzzy. Besides, designing a routing algorithm in the ECNUM needs to consider multiple network parameters to satisfy various QoS requirements, which is a multiple input single output (MISO) problem. It is difficult to define an accurate mathematical expression under the multiple constraints. Fuzzy decision theory (FDT) has potential for dealing with the MISO problem, which can calculate the results through heuristic reasoning instead of establishing complex and accurate mathematical models.⁶ Therefore, FDT is suitable to solve routing decision problems with multiple QoS constraints.

The focus of this study is to design routing protocols for emergency communication networks in underground mines. We have analyzed transmission characteristics of emergency data and regular data and chosen appropriate network parameters according to the data type to implement our work. Based on FDT, this article has proposed the fuzzy-logic-based data-differentiated service supported routing protocol (FDDSP) to satisfy different data transmission requirements in the ECNUM and optimize network performance. In addition, the path soft handoff strategy is proposed to maintain the continuity of the data transmission and improve the QoS when the path performance is degenerated.

The remainder of this article is organized as follows. Section “Related work” reviews the related work. Section “Preliminaries” introduces the applicability of FDT in the WMNs routing decision. Section “FDDSP” proposes FDDSP and describes it in detail. Section “Simulation experiment” presents the results of the simulation experiments. Finally, the conclusion of the study is presented in section “Conclusion.”

Related work

China is a major producer and consumer of coal. Coal consumption accounts for 70% of the non-renewable energy consumption at the present stage.⁷ In recent years, the safety of coal mining has been paid more attention with the increase in depth and intensity in production processes. The terrain of an underground mine is complex, and the environment of mining production is terrible. Traditional wired networks cannot provide reliable communication services.⁸ To improve the robustness of the communication networks in underground mines, a wired network generally forms the backbone network, and the wireless networks work as a complement. There are some wireless technologies used in mine communications, as shown in Table 1.

Table 1.

Mine wireless communication system.

Wireless communication system	Defect
Very-low-frequency (VLF) through-the-ground communication system	Small channel capacity, susceptible to electromagnetic interference, and cumbersome system deployment
Middle-frequency(MF) inductive communication system	Special induction cable required and short induction distance
Leakage communication system	Leakage coaxial cable required, limited coverage, and poor anti-interference capability
Cellular mobile communication system	Expensive equipment and complex deployment
PHS wireless communication system	Narrowband communication and non-support multimedia service
Wi-Fi communication system	Poor scalability

The communication systems listed in Table 1 are strongly dependent on the network infrastructures. When a disaster occurs, the damage to infrastructures may lead to the paralysis of communications in the underground mine. HWMNs are different from traditional wireless networks due to their abilities of self-organization, self-healing, and extensibility. Besides, HWMNs can support the high capacity of network transmissions, efficient broadband access, and reliable multimedia support. Therefore, the HWMN can be used to build the ECNUM.^9–12

With the process of coal mining and tunnel excavation, the tunnels extend gradually and form a hundreds of, or even a thousands of, meters long strip structure. Therefore, HWMNs with long-chain topologies in underground mines can only transfer information through multiple hops. For multi-hop networks, the hop count of a path directly affects the transmission delay and data delivery ratio. The hop count should be reduced as much as possible since an overlong path will lead to unstable data transmissions.

The MCs are powered by limited capacity batteries, which directly affect the lifetime of MCs and the ECNUM.^13,14 In emergency communications, the energy supply in the affected area is interrupted. The energy of nodes cannot be supplemented, which will make the problem of energy limitations more prominent.¹⁵ The network infrastructure, such as the MRs in the disaster area, may be partially or totally damaged. As a result, most of the network load is transmitted by MCs. Therefore, the energy balance should be achieved to prolong the network lifetime. In addition, the computing power and storage capacity of nodes are limited. An unbalanced load allocation can cause an overflow of the buffer queue and an acceleration of energy consumption. Therefore, the load balance should also be considered in the design of the routing algorithm for the ECNUM.

The necessity of using multiple criteria in network routing decisions is proposed in the study by Zhang and Long.¹⁶ Basarkod and Manvi¹⁷ implemented a routing protocol that can avoid congestion and adapt to frequent topological changes by using node mobility (measured by the node stability model) and network congestion (determined by the channel load and buffer occupancy) as routing metrics. A method of measuring link quality is proposed in the study by Jabbar et al.¹⁸ using the idle time, residual energy, and buffer space of nodes. Ma and Denko¹⁹ used two parameters, the cumulative expected transmission time and residual buffer, to choose the best path to achieve load balance. Lu et al.²⁰ proposed a multi-criteria routing metric (MRM) based on the network delay, hop count, and link load. MRM divided data into two types: urgent and non-urgent. End-to-end delay was calculated when transmitting urgent data, and hop count and link load were measured when transmitting non-urgent data. The aforementioned studies implemented multi-metric routing decisions by introducing proportional adjustment factors to build a multi-parameter adjustment model (MPAM). However, there is no direct quantitative relationship between these parameters. The evaluation of the path is the fuzzy process. It is difficult to give an accurate mathematical description of the MPAM.

FDT, which was proposed by Zadeh in 1965, is widely used in engineering research and industrial controls.^21–23 FDT provides a new solution for the MPAM control system as well as routing decisions constrained by the multiple parameters of the network.²⁴ Marimuthu and Kannammal²⁵ discussed the selection of network parameters when applying FDT to the routing decision of ad hoc networks. Different fuzzy routing decision methods were proposed according to the type of protocol (proactive or reactive) in the study by Ghazisaidi et al.²⁶ The proactive routing protocol considered the hop count, expected transmission count (ETX), and congestion, while the reactive routing protocol considered the delay, hop count, signal strength, and congestion. Chelliah et al.²⁷ focused their efforts on trying to determine a fuzzy multi-constraint ad hoc on-demand distance vector (AODV) routing. The method made routing decisions based on more than one constraint such as buffer occupancy, node energy, and hop count. Torshiz et al.²⁸ proposed fuzzy energy-based AODV (FE-AODV). FE-AODV chose three parameters, the min-bandwidth, hop count and battery life, as inputs of the fuzzy decision system (FDS). However, FE-AODV needs to collect multiple path information and will cause additional overhead.

Abbas et al.²⁹ proposed a new method called FUZZY-AODV to optimize the AODV protocol. FUZZY-AODV had three inputs, including the residual energy, movement speed, and hop count. FUZZY-AODV reduced the possibility of a link break and improved the stability and lifetime of the network by selecting the next hop node with a higher trust value (low mobility, small hop count, and sufficient energy). However, FUZZY-AODV did not consider the link load and may cause congestion. Some clustering routing protocols were proposed in previous studies.^30–32 These protocols used FDT to select the cluster head and improved the performance of the network. The aforementioned study shows that FDT is very suited for implementing multi-parameter routing.

However, the data types and transmitting requirements are more complex in ECNUM than general applications. In this article, based on the analysis of the transmission characteristics of emergency data and regular data, the multi-parameter routing model is constructed. FDDSP is proposed to satisfy different data transmission requirements in the ECNUM and optimize the network performance.

Preliminaries

Data description

Data transmitted in ECNUM can be divided into regular data and emergency data according to their sources, amounts, and transmission priorities.³³ Regular data include underground environmental information (such as gas concentrations, wind speed, temperature, and other monitoring data), personnel locations, and video streams. Emergency data are produced by the unstable environmental factors or unexpected situations after a disaster and mainly come from the early warning or alarm, such as gas leakage, rock permeability, and abnormal vital signs. The comparison of emergency data and regular data is shown in Table 2.

Table 2.

Comparison of emergency data and regular data.

Data type	Emergency data	Regular data
Data amount	Small	Large
Data priority	High	Low
Queue waiting	No	Yes
Energy consumption	Small	Large
Throughput demand	Small	Large
Delay tolerance	Low	High
Reliability demand	High	Low
Optimization goal	Reliable and fast transmission	High throughput and long network lifetime

As shown in the comparison, the transmission of regular data requires higher throughput, balanced energy consumption, and load pressure. However, regular data do not require real-time transmission and are tolerant of a certain degree of packet loss. Emergency data have strong bursts and small data sizes. Their transmission requires higher real-time performance and reliability, but without a large throughput. When there are both emergency data and regular data, the emergency data will be pushed to the front of the buffer queue and sent first because of their higher priority. If there are multiple emergency packets, they will be sent in the order of first-in-first-out (FIFO). Regular data are sent in the normal FIFO order.

Fuzzy decision model

FDT is used to achieve multi-QoS constrained routing decisions in this article. The FDS mainly contains the fuzzifier, fuzzy inference engine, and defuzzifier. First, the fuzzifier transforms the crisp inputs into linguistic values through membership functions. Then, the fuzzy inference engine, which contains a rule base (a series of IF-THEN rules) and reasoning methods, will execute the reasoning computation with linguistic variables obtained by fuzzification.³⁴ Finally, the defuzzifier transforms the results into crisp outputs to make decisions.

The FDS for emergency data is shown in Figure 3. The crisp inputs are the hop count (HC), path transmission quality (PTQ), and path residual energy (PRE). HC indicates the real-time and stability performance of the path. The PTQ and PRE represent the reliability of data transmissions and path lifetime, respectively. The path trust value (PTV), as the routing metric, is the crisp output of the FDS.

Figure 3.

FDS of emergency data.

The FDS for regular data chooses the PRE, HC, and path buffer quality (PBQ) as crisp inputs. The crisp output is the PTV. The PBQ indicates the load balance of the path. The main variables used in this article are shown in Table 3.

Table 3.

Variables table.

N	The number of nodes contained in the path.
$PRR (i, w)$	Packet delivery ratio of node i in time window w.
$e_{i, t}$	The percentage of residual energy of node i at time t.
$E_{res} (i, t)$	The residual energy of node i at time t.
$E_{init} (i)$	Initial energy of node i.
$b_{i, t}$	The percentage of the residual buffer of node i at time t.
$B_{res} (i, t)$	The residual buffer of node i at time t.
$B_{all} (i)$	Total buffer of node i.
$B_{μ} (t)$	The average of the residual buffer of all nodes in the path at time t.
$B_{σ} (t)$	The standard deviation of the residual buffer of all nodes in the path at time t.
HC	Hop count of the path.
PTQ	Transmission quality of the path.
PRE	Residual energy of the path.
PBQ	Buffer quality of the path.
PTV	Trust value of the path.

FDDSP

FDS for emergency data

The transmission of emergency data requires low delay and high reliability to alarm the danger quickly. Three parameters are set in an emergency FDS: HC, PTQ, and PRE.

HC

The delay of data transmission mainly comes from the time consumed with node forwarding and queue waiting. Emergency data have higher priority and are forwarded directly by nodes instead of being added to the end of the waiting queue. Therefore, the transmission delay of emergency data is mainly related to the HC. Besides, the stability of path is greatly influenced by HC. The path with the lower HC should have more opportunities to be selected as the routing path.

PTQ

Because wireless communications are unstable, the quality of intermediate links directly affects the data transmission. Some link quality estimation (LQE) methods are proposed in the study by Baccour et al.³⁵ It is difficult for nodes in the ECNUM to support the complex LQE algorithm because of its limited energy and computing power. In this article, the packet reception ratio (PRR) is used to measure the link quality. The calculation of PRR is shown as follows

P R R (i, w) = \frac{N u m b e r o f r e c e i v e d p a c k e t s}{N u m b e r o f s e n t p a c k e t s}

(1)

The range of $PRR (i, w)$ is [0, 1]. To achieve successful data transmission, it must be ensured that every intermediate node transmits data successfully.³⁶ Therefore, the PTQ of the path can be defined as follows

PTQ = Π_{i = 1}^{i = N} PRR (i, w)

(2)

where PTQ is the multiplication of the quality of all links contained in the path. The value range of PTQ is [0, 1].

PRE

Because of the serious energy limitations after disasters, it is necessary to extend the network lifetime as far as possible. The percentage of residual energy in node i at time t is shown as follows

e_{i, t} = \frac{E_{res} (i, t)}{E_{init} (i)}

(3)

The survival state of a path is constrained by the residual energy of the intermediate nodes. If an intermediate node dies, the path will break. Therefore, the PRE is defined as follows

PRE = min (e_{i, t}) i = {1, 2, \dots, N}

(4)

The HC, PTQ, and PRE are inputs of the emergency FDS as shown in Figure 3, and the PTV is the output. Figure 4 describes the membership functions of inputs and output. The FDS contains 27 rules in its rule base and two examples are shown as follows:

If (HC = CLOSE && PRE = sufficient && PTQ = HIGH), then PTV = S (4), and

If (HC = FAR && PRE = SHORT && PTQ = LOW), then PTV = W (4).

Figure 4.

Membership functions of an emergency FDS: (a) HC: hop count, (b) PRE: path residual energy, (c) PTQ: path transmission quality, and (d) PTV: path trust value.

The value of the PTV is calculated by the process of the defuzzifier. The path with largest PTV will be chosen to transmit the data. The centroid defuzzification method is used in this article and the mathematical expression is as follows³⁷

PTV = \frac{\sum μ_{A} (x) \cdot x}{\sum μ_{A} (x)}

(5)

FDS for regular data

The amount of regular data is large, and it will persist in the whole lifetime of the ECNUM. The energy limitations after a disaster make the energy balance of nodes more important. In addition, the transmission of regular data demands high throughput, which requires the implementation of load balance among nodes to improve the service quality of the network. Three parameters are set in regular FDS: PRE, PBQ, and HC.

The PRE directly affects the path stability and network lifetime. For the large amount of regular data transmissions, it can effectively reduce the energy holes in the network when choosing paths with larger PRE.

Compared with emergency data, the transmission of regular data has fewer limitations for delays. Regular data will join in the waiting queue when the load of the transmitting node is large. The PBQ reflects the amount of the path residual buffer (PRB) and the equilibrium degree of the used buffer. The residual buffer ratio of intermediate nodes is defined as follows

b_{i, t} = \frac{B_{res} (i, t)}{B_{all} (i)}

(6)

where the value range of $bi, t$ is [0, 1]. The transmission capacity of the path is constrained by the weakest intermediate nodes. The PRB is defined as the minimum value of $bi, t$ in the path

PRB = min (b_{i, t}) i = {1, 2, \dots, N}

(7)

The value range of the PRB is [0, 1].

The balance degree of the residual buffer in the path can be expressed by the standard deviation $B σ (t)$ as follows

B_{σ} (t) = \sqrt{\frac{1}{N} \sum_{i = 1}^{N} {(b_{i, t} - B_{μ} (t))}^{2}}

(8)

where

B_{μ} (t) = \frac{\sum_{i = 1}^{N} b_{i, t}}{n}

(9)

pbq is defined as follows

pbq = PRB - B_{σ} (t)

(10)

Because the range of $B_{σ} (t)$ is [0, 0.5], the range of pbq [–0.5, 1] can be derived. To facilitate its description in the FDS, pbq is normalized as follows

PBQ = \frac{pbq + 0.5}{1.5}

(11)

When the value of PRB is large and $B_{σ} (t)$ is small, PBQ has a large value, which indicates that the path that is undertaken has a low and balanced load.

The membership functions of the regular FDS are shown in Figure 5. There are 27 rules in the rule base and some examples are shown as follows:

If (HC = CLOSE && PRE = SUFFICIENT && PBQ = HIGH), then PTV = S (4), and

If (HC = FAR && PRE = SHORT && PBQ = LOW), then PTV = W (4).

Figure 5.

Membership functions of the regular FDS: (a) HC: hop count, (b) PRE: path residual energy, (c) PBQ: path buffer quality, and (d) PTV: path trust value.

Implementation of FDDSP

Algorithm process

To distinguish the data types, FDDSP adds the data type (DT) field to the route request (RREQ) packets. There are also other fields, such as HC, PTQ, PRE, and PBQ, that store the corresponding parameters mentioned in sections “FDS for emergency data” and “FDS for regular data.” These fields will be inputted into the FDS when making routing decisions. The structure of a RREQ is shown in Table 4.

Table 4.

Structure of an RREQ.

Type	Flag		DT
HC	PTQ	PRE	PBQ
Destination IP address
Destination sequence number
Source IP address
Source sequence number

DT: data type; HC: hop count; PTQ: path transmission quality; PRE: path residual energy; PBQ: path buffer quality.

FDDSP fills the DT field according to formula (12) and initiates the routing requests

DT = {\begin{matrix} 1 emergency data \\ 0 regular data \end{matrix}

(12)

FDDSP will select different FDSs according to the value of DT when receiving an RREQ. The path selection process is shown in Figure 6.

Figure 6.

Path selection process of FDDSP.

FDDSP may choose different transmission paths to transmit different types of data, even though the data are coming from the same source. To distinguish the path as being for emergency data or regular data, the routing type (RTT) field is added to the routing table. The value of RTT is defined as follows

RTT = {\begin{matrix} 1 DT = 1 \\ 0 DT = 0 \end{matrix}

(13)

The structure of the routing table is shown in Table 5.

Table 5.

Routing table.

Next hop	RTT	PTV
Destination IP address
Destination sequence number

RTT: routing type; PTV: path trust value.

The PTV field is used to store the output of the FDS in sections “FDS for emergency data ” and “FDS for regular data.” PTV is used as the routing metric and is updated as the following rule

P T V = {\begin{matrix} c u r r e n t_P T V c u r r e n t_P T V \geq n e w_P T V \\ n e w_P T V c u r r e n t_P T V < n e w_P T V \end{matrix}

(14)

where $current_PTV$ represents the current value of PTV and $new_PTV$ represents the new value of PTV, which is calculated from the parameters in the RREQ. FDDSP will update the routing table according to the RTT and PTV and will choose the path with the largest PTV when RTT has the same value.

Soft handoff strategy

The quality of the path will decrease and the value of PTV will change with the running of the network. To ensure that the energy balance and loads are proportionate, there are two trigger conditions to start the path handoff process for node i at time t.

$e_{i, t}$ has decreased by 30% from the time of the path’s establishment to now.

The node has a heavy load.

Due to the fluctuation of the node loads, in order to avoid frequent path switching, it is necessary to evaluate the load condition and divide the node loads into light loads and heavy loads, as shown in Figure 7. The buffer occupancy $(1 - b_{i, t})$ is obtained periodically and the period is set to 1 s. At the same time, a Threshold (the value is 75%) is set in FDDSP. FDDSP sets a time interval (10 s) as the load evaluation time window (TW). If the time that is satisfied $(1 - b_{i, t} > Threshold)$ is longer than 70% of the TW, the load is heavy. Otherwise, the node carries a light load.

Figure 7.

Load determination curve.

When a path requires a handover, node i will send a path rebuild request to the source node S, but the original transmission path is still active, as shown in Figure 8(a) and (b). When S receives the request, a new path construction process will be started to build a spare path, as shown in Figure 8(c). When the new path is ready, the data will be transferred to the new path. A seamless soft handoff has been achieved, as shown in Figure 8(d).

Figure 8.

Process of a path soft handoff: (a) original path, (b) path rebuild request, (c) new path construction, and (d) path soft handoff.

Simulation experiment

Simulation environment and experimental parameters

Based on the NS2 simulation platform, the performance of FDDSP has been tested on average end-to-end delay, delivery ratio, throughput, and network lifetime, which are compared with AODV,³⁸ MRM,²⁰ and FUZZY-AODV.²⁹ The performance indicators are defined as follows.

Average end-to-end delay

The average value of constant bitrate (CBR) packets delays from each source node to the gateway node. The average end-to-end delay can be defined as follows

Avg_Delay = \frac{\sum_{i}^{I} CBR (i)}{I}

(15)

where $CBR (i)$ is the delay of packet i and I is the number of packets in an experimental process.

Delivery ratio

Delivery ratio is the ratio of CBR packets arriving at the gateway node to the total packets sent by source nodes. The delivery ratio can be defined as follows

D e l i v e r y r a t i o = \frac{P a c k e t s r e c e i v e d b y g a t e w a y}{A l l p a c k e t s}

(16)

Throughput

Throughput is the amount of data received by the gateway node in 1 s.

Network lifetime

The time when the first node dies is defined as the network lifetime.

The parameters of the simulation network are listed in Table 6.

Table 6.

Simulation parameters.

Topographic size	1000 × 20 m²
Simulation time	1500 s
Transmission range	200 m
Traffic type	Constant bitrate (CBR)
Regular packet size	512 bytes
Regular data flow rate	0.6 Mbps
Emergency packet size	64 bytes
Emergency data flow rate	0.1 Mbps
Number of CBR flows	From 2 to 20
Gateways	1
Mesh routers	7
Mesh clients (MCs)	93
Initial energy of MC	100 J
Transmission power	0.660 W
Receiving power	0.395 W
Idle power	0.035 W

The network has been set in a long and narrow area of 1000 × 20 m² to simulate the network scene in Figure 1 and deploy seven MRs, one gateway, and the number of MCs is [50,150] in the network. Different packet sizes and data flow rates are set to simulate the different amounts of data. The packet size of emergency data is 64 bytes, and its data flow rate is 0.1 Mbps. The size of regular data is 512 bytes, and its data flow rate is 0.6 Mbps. The experiment sets three transmission scenarios: emergency data transmissions, regular data transmissions, and mixed data (emergency data and regular data coexisting in the network) transmissions. With a fixed network size (MRs = 7, MCs = 93), the performances of the aforementioned protocols in the three scenarios are tested under different load pressures by increasing the number of CBR data streams (from 2 to 20). Each data point is the average of 20 experiments.

Besides, we test average end-to-end delay and network lifetime with different network sizes (MRs = 2, MCs from 50 to 150). Each point is the average of 20 experiments. In every 20 experiments, we fix the number of MCs (such as MCs = 50 or MCs = 60) and change the number of CBR streams (1–20 CBRs).

Performance analysis

Delivery ratio

The delivery ratios of protocols in three data transmission scenarios are compared in Figures 9 –11. Regardless of whether emergency data, regular data, or mixed data exist, the delivery ratio of FDDSP is the highest, followed by the FUZZY-AODV and MRM, and AODV is the lowest. AODV selects the path with the lowest hop count to transmit data but cannot avoid the congestion or unbalanced energy consumption. When the path is broken or the transmission quality is substantially reduced, a new path request will be initiated by AODV. The path rebuild strategy of AODV causes time waste and packet loss. FUZZY-AODV considers the residual energy and hop count to select the path and can reduce the packet loss caused by node death. However, it is not sensitive to the congestion because the load factor is not involved in the routing metric. The transmission of emergency data in MRM only considers the delay. This single routing metric makes MRM unable to efficiently adjust the routing, thus resulting in some packet losses. MRM considers the hop count and link load when transmitting regular data. However, the metric does not contain the energy factor and cannot avoid the excessive use of MCs with low residual energy. Therefore, some packets are lost due to the death of nodes. When transmitting emergency data, FDDSP takes PTQ and PRE as fuzzy inputs to avoid choosing the path with low transmission quality and less residual energy. When selecting the path for regular data, the PBQ and PRE are used as fuzzy inputs to avoid the packet loss caused by congestion and node death. In addition, when the path quality changes, the path soft handoff can maintain the continuity of data transmission and choose a new path to reduce the packet loss further.

Figure 9.

Delivery ratio of emergency data.

Figure 10.

Delivery ratio of regular data.

Figure 11.

Delivery ratio of mixed data.

The delivery ratios of FDDSP in three data transmission scenarios are shown in Figure 12. The results show that the emergency data transmission has a higher delivery ratio than that of the other two data transmission scenarios. There are two main reasons for this result: (1) the amount of emergency data is small and (2) the optimization goals of the transmission of emergency or regular data are different. The transmission of emergency data emphasizes the reliability, while regular data can tolerate a small amount of packet losses.

Figure 12.

Delivery ratio in different scenarios.

Average end-to-end delay

Figures 13 –18 show the comparisons of end-to-end delay in three scenarios. Whether the network size is fixed or not, FDDSP has the lowest delay. When the network size changes, the delay decreases with the number of nodes increasing. The path selection mechanism of AODV cannot avoid the heavy load or congestion nodes effectively. In addition, AODV may choose low-energy nodes to transmit data, which results in the premature death of nodes as well as path breaking and reconstruction. Therefore, AODV has the highest end-to-end delay, especially under high loads. FUZZY-AODV considers the residual energy to avoid choosing the low-energy nodes. Therefore, it can reduce the delay caused by a broken path when the load is heavy. However, the node’s current load pressure is not considered (e.g., the node with sufficient residual energy may carry a heavy load), and thus FUZZY-AODV may still cause delays due to congestion. MRM selects the path with the lowest delay when transmitting emergency data. When the load is light, MRM has lowest end-to-end delay. However, it cannot effectively avoid the delays caused by congestion and cannot change the transmission path in time when the load pressure is increased. The transmission of regular data can effectively avoid the use of heavy load nodes in MRM, but the delay caused by node early death cannot be solved because of using a low-energy node. FDDSP can send emergency data first without waiting. In addition, FDDSP uses the emergency FDS to choose a path with a high delivery ratio and short length. Therefore, FDDSP has lower end-to-end delay when transmitting emergency data. When transmitting regular data, FDDSP can avoid the congestion area and reduce the delay efficiently through the use of the regular FDS.

Figure 13.

End-to-end delay of emergency data versus CBR.

Figure 14.

End-to-end delay of regular data versus CBR.

Figure 15.

End-to-end delay of mixed data versus CBR.

Figure 16.

End-to-end delay of emergency data versus MCs.

Figure 17.

End-to-end delay of regular data versus MCs.

Figure 18.

End-to-end delay of mixed data versus MCs.

The end-to-end delays of FDDSP in the three data transmission scenarios are shown in Figures 19 and 20. The transmission of the emergency data has minimal end-to-end delay, which meets the real-time requirement. While the transmission of regular data tolerates a certain delay, the results show that it is still within the acceptable range.

Figure 19.

End-to-end delay in different scenarios versus CBR.

Figure 20.

End-to-end delay in different scenarios versus MCs.

Throughput

As shown in Figures 21 –23, FDDSP has the highest throughput, followed by FUZZY-AODV and MRM, and AODV has the lowest. From the analysis of sections “Delivery ratio” and “Average end-to-end delay,” it can be concluded that high latency and more packet loss result in lower throughput in AODV. FUZZY-AODV can reduce the network overhead caused by path breaking, and thus, its throughput is higher than that of AODV. MRM considers the path load in routing decisions, and thus, it can effectively avoid congestion when transmitting regular data. However, the reduction in the throughput has not been effectively prevented, which is caused by link breakage and path reconstruction. FDDSP has taken full account of the energy and load of the path and can reduce the frequency of path breakage and reconstruction. In addition, the soft handoff strategy of FDDSP can maintain the continuity of data transmission. Therefore, FDDSP has higher throughput than that of other similar protocols.

Figure 21.

Throughput of emergency data.

Figure 22.

Throughput of regular data.

Figure 23.

Throughput of mixed data.

Network lifetime

Whether the network size is fixed or not, FDDSP has the longest network lifetime. The network lifetime extends as the network scale increases. FDDSP and FUZZY-AODV consider the residual energy as the routing metric to achieve the energy balance, and thus, their lifetimes are longer than those of AODV and MRM. The regular FDS of FDDSP can distribute the network load more evenly than FUZZY-AODV can. Therefore, FDDSP has a longer lifetime than the FUZZY-AODV does. In addition, the path soft handover strategy of FDDSP can detect and switch off low-quality paths in time, which can avoid overusing paths with heavy loads and prolong the network lifetime (Figures 24 –29).

Figure 24.

Network lifetime of emergency data versus CBR.

Figure 25.

Network lifetime of regular data versus CBR.

Figure 26.

Network lifetime of mixed data versus CBR.

Figure 27.

Network lifetime of emergency data versus MCs.

Figure 28.

Network lifetime of regular data versus MCs.

Figure 29.

Network lifetime of mixed data versus MCs.

Conclusion

It is necessary to build the ECNUM on time to accelerate rescue progress after a disaster. There are some special problems in the ECNUM, such as different data types, the destruction of infrastructure, and the limited energy of nodes. To satisfy different data transmission requirements and solve the post-disaster problems, this article has solved the routing decision problem with multiple QoS constraints based on FDT and the proposed FDDSP. In this article, the data types are divided into emergency data and regular data. Emergency data require real-time and reliable transmissions. Regular data require a high throughput and balanced energy consumption. FDDSP chooses different FDSs to make routing decisions for different types of data and provides differentiated data services to optimize the transmission quality. In addition, the soft handoff strategy of FDDSP has achieved the reliability and continuity of data transmissions. The performances of FDDSP in three data transmission scenarios under different load pressures and network size are simulated and compared with those of AODV, FUZZY-AODV, and MRM. The results show that FDDSP has the best performance in end-to-end delays and the delivery ratio when transmitting emergency data. When transmitting regular data, FDDSP can achieve high throughput on the basis of guaranteeing fewer end-to-end delays and a higher delivery ratio.

To ensure the personal safety in the affected area and avoid secondary damage, the mobility of client nodes held by miners is assumed to be low in this article. However, in practical applications, node movement is inevitable, which will cause unstable network topology and affect the network’s performance. In future work, more attention will be paid to the node mobility in the ECNUM to improve the network performance under real application scenarios in underground mines.

Footnotes

Handling Editor: Sergio Toral

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: The financial support for this work provided by the Fundamental Research Funds for the Central Universities (no. 2017XKQY077) is gratefully acknowledged.

ORCID iD

Haifeng Jiang

References

Reddy

Saketh

Dhar

. Review of sensor technology for mine safety monitoring systems: a holistic approach. In: IEEE first international conference on control, measurement and instrumentation, Kolkata, India, 8–10 January 2016, pp.429–434. New York: IEEE.

Zhang

Liu

Ding

. Study on topology reconstruction and data transmission of mine emergency communication network. Coal Sci Technol 2015; 43: 95–100.

Sharma

Kumar

Singh

. Routing in wireless mesh networks: three new nature inspired approaches. Wirel Pers Commun 2015; 83: 3157–3179.

Jiang

Qian

Sun

et al . Energy optimal routing for long chain-type wireless sensor networks in underground mines. Int J Min Sci Technol 2011; 21(1): 17–21.

Novak

Snyder

Kohler

. Postaccident mine communications and tracking systems. IEEE T Ind Appl 2010; 46(2): 712–719.

Haider

Yusuf

. A fuzzy approach to energy optimized routing for wireless sensor networks. Int Arab J Inf Techn 2007; 6(2): 179–185.

Jiang

Sun

. Energy optimized routing algorithm for hybrid wireless mesh networks in coal mine. Int J Distrib Sens N 2015; 11, https://journals.sagepub.com/doi/full/10.1155/2015/237697.

Liu

. Underground coal mine monitoring with wireless sensor networks. ACM T Sensor Network 2009; 5(2): 10.

Odabasi

Zaim

. A survey on wireless mesh networks, routing metrics and protocols. Int J Electron Mechan Mechatron Eng 2011; 2(1): 92–104.

10.

Song

Wang

Dai

. A emergency communication system based on WMN in underground mine. In: International conference on computer application and system modeling, Taiyun, China, 22–24 October 2010, pp.V4-624–V4-627. New York: IEEE.

11.

Sicignano

Tardioli

Cabrero

et al . Real-time wireless multi-hop protocol in underground voice communication. Ad Hoc Netw 2013; 11(4): 1484–1496.

12.

Yang

Zhang

Liu

. Constructing of wireless emergency communication system for underground coal mine based on WMN technology. J Coal Sci Eng 2010; 16(4): 441–448.

13.

Kai

. A mine-environment-based energy-efficient routing algorithm for wireless sensor network. In: International conference on hybrid intelligent systems, Shenyang, China, 12–14 August 2009, pp.215–218. New York: IEEE.

14.

Youssef

Ibrahim

Abdelatif

et al . Routing metrics of cognitive radio networks: a survey. IEEE Commun Surv Tutor 2015; 16(1): 92–109.

15.

Jafarian

Jaseemuddin

. Routing of emergency data in a wireless sensor network for mines. In: IEEE international conference on communications, Beijing, China, 19–23 May 2008, pp.2813–2818. New York: IEEE.

16.

Zhang

Long

. A fuzzy routing mechanism in next-generation networks. In: IASTED international conference on intelligent systems and control, Tsukuba, Japan, 1–4 October, 2002, pp.86–91. ACTA Press.

17.

Basarkod

Manvi

. Node movement stability and congestion aware anycast routing in mobile ad hoc networks. In: 2014 IEEE international advance computing conference (IACC), Gurgaon, India, 21–22 February 2014, pp.124–131. New York: IEEE.

18.

Jabbar

Ismail

Norton

. Multi-criteria based multipath OLSR for battery and queue-aware routing in multi-hop ad hoc wireless networks. Wirel Netw 2014; 21(4): 1–18.

19.

Denko

. A routing metric for load-balancing in wireless mesh networks. In: International conference on advanced information networking and applications workshops, Niagara Falls, ON, Canada, 21–23 May 2007, pp.409–414. New York: IEEE.

20.

Jiang

Han

et al . Multi-criteria routing metric for supporting data-differentiated service in hybrid wireless mesh networks in coal mines. Int J Distrib Sens N 2017; 13(1): 1550147716689796.

21.

Zadeh

. Fuzzy sets. Inform Contr 1965; 8(3): 338–353.

22.

Huang

Gao

et al . Stability analysis of the simplest Takagi-Sugeno fuzzy control system using circle criterion. J Syst Eng Electron 2007; 177(2): 4387–4409.

23.

Vasilakos

Saltouros

Atlassis

et al . Optimizing QoS routing in hierarchical ATM networks using computational intelligence techniques. IEEE T Syst Man Cy C 2003; 33(3): 297–312.

24.

Bellman

Zadeh

. Decision-making in a fuzzy environment. Manage Sci 1970; 17(4): B141–B164.

25.

Marimuthu

Kannammal

. A survey on fuzzy based QoS routing in Mobile Ad Hoc Networks. In: International conference on intelligent systems and control, Coimbatore, India, 4–5 January 2013, pp.376–379. New York: IEEE.

26.

Ghazisaidi

Assi

Maier

. Intelligent wireless mesh path selection algorithm using fuzzy decision making. Wirel Netw 2012; 18(2): 129–146.

27.

Chelliah

Sankaran

Prasad

et al . Routing for wireless mesh networks with multiple constraints using fuzzy logic. Int Arab J Inf Techn 2011; 9(1): 1.

28.

Torshiz

Amintoosi

Movaghar

. A fuzzy energy-based extension to AODV routing. In: International symposium on telecommunications, Tehran, Iran, 27–28 August 2008, pp.371–375. New York: IEEE.

29.

Abbas

Ilkan

Ozen

. Fuzzy approach to improving route stability of the AODV routing protocol. EURASIP J Wirel Comm 2015; 2015(1): 235.

30.

Maurya

Daniel

. An energy efficient routing protocol under distance, energy and load parameter for heterogeneous wireless sensor networks. In: International conference on information technology, Bhubaneswar, India, 22–24 December 2014, pp.161–166. New York: IEEE.

31.

Messaoudi

Elkamel

Helali

et al . Distributed fuzzy logic based routing protocol for wireless sensor networks. In: International conference on software, telecommunications and computer networks, Split, 22–24 September 2016, pp.1–7. New York: IEEE.

32.

Yusuf

Haider

. Energy-aware fuzzy routing for wireless sensor networks. In: Proceedings of the IEEE symposium on emerging technologies, Islamabad, Pakistan, 18 September 2005, pp.63–69. New York: IEEE.

33.

Wenfeng

Jie

Peng

. Mine multimedia emergency communication system. In: 2007 international conference on wireless communications, networking and mobile computing, Shanghai, China, 21–25 September 2007. New York: IEEE.

34.

Dernoncourt

. Introduction to fuzzy logic. Cambridge, MA: Massachusetts Institute of Technology, 2013, p.21.

35.

Baccour

Koubaa

Ben Jamaa

et al . A comparative simulation study of link quality estimators in wireless sensor networks. In: IEEE international symposium on modeling, analysis & simulation of computer and telecommunication systems, London, 21–23 September 2009, pp.1–10. New York: IEEE.

36.

Yang

Zhang

Gao

et al . Research of HS-AODV routing protocol based on fuzzy control theory. In: Chinese automation congress, Wuhan, China, 27–29 November 2015, pp.1023–1028. New York: IEEE.

37.

Nurcahyo

Shamsuddin

Alias

et al . Selection of defuzzification method to obtain crisp value for representing uncertain data in a modified sweep algorithm. J Comp Sci Technol 2003; 3: 22–28.

38.

Perkins

Royer

. Ad-hoc on-demand distance vector routing. In: The workshop on mobile computing systems & applications, New Orleans, LA, 25–26 February 1999, pp.94–95. New York: IEEE.

Fuzzy-logic-based data-differentiated service supported routing protocol for emergency communication networks in underground mines

Abstract

Keywords

Introduction

Related work

Preliminaries

Data description

Fuzzy decision model

FDDSP

FDS for emergency data

HC

PTQ

PRE

FDS for regular data

Implementation of FDDSP

Algorithm process

Soft handoff strategy

Simulation experiment

Simulation environment and experimental parameters

Average end-to-end delay

Delivery ratio

Throughput

Network lifetime

Performance analysis

Delivery ratio

Average end-to-end delay

Throughput

Network lifetime

Conclusion

Footnotes

Declaration of conflicting interests

Funding

ORCID iD

References