Lifetime Forecast Routing With Node Mobility

Published Date: 02 Nov 2017

CHAPTER 3

3.1 OVERVIEW

A mobile ad hoc network is a movable, multi-hop wireless network by no fixed infrastructure. Dynamic topologies because of mobility and limited bandwidth and battery power build the routing difficulty in ad hoc networks more demanding than conventional wired networks. A key to develop efficient routing protocols for such networks lies in keeping the routing overhead negligible. A novel category of on-demand routing protocols such as DSR, AODV, TORA try to decrease routing overhead through only maintaining routes among nodes taking part in data communication.

Energy aware and link stable paths turn out to be key issues in designing scalable routing protocols in Mobile ad hoc networks (MANETs). The objective of this work is the proposal of a new routing model which could be able to account for a joint metric of more link stability and less energy consumption in MANET.

Energy is a significant resource that needs to be preserved in order to extend the lifetime of the network. In contrast, the link and path stability among nodes permits the diminution of control overhead and could offer some benefits also in terms of energy saving over MANET. However, as will be shown in this work, the choice of more stable routes below nodes mobility could direct to the choice of shorter routes. This is not forever appropriate in terms of energy consumption. Conversely, on occasion, attempting to optimize the energy could lead to the choice of more weak routes. Therefore, it is obvious that both the abovementioned parameters (specifically, link stability linked with the nodes mobility and energy consumption) must be measured in designing routing protocols, which permit right tradeoff between route stability and least energy consumption to be attained.

Fig. 3.1 Objective of proposed model

The main aspire of this work is to propose a routing model within a MANET. As shown in fig. 3.1, the model attempts to reduce concurrently the energy consumption of the mobile nodes and exploit the link stability of the broadcasts, when selecting paths for individual broadcasts.

The thought of considering, simultaneously, energy consumption and link stability is provoked by the surveillance that most routing protocols tend to choose shorter routes, in this way high effectiveness in using wireless bandwidth and augment path stability are ensured. On the other hand, such routes may endure from higher energy consumption, because higher broadcasts ranges are wanted. Consequently, in order to take into account the energy consumption and link stability of mobile nodes, Node Mobility and Density Classifier Model is proposed.

To handle dynamic topology of MANET with ZRP protocol, Lifetime forecast Routing (LFR) is designed. On the basis of Lifetime Forecast Routing (LFR) with Node Mobility and Density Classifier (NMDC) model presented, a routing protocol is proposed and its validity is experimentally investigated through simulations. A comparison with other known approaches, such as Link-stAbility and Energy aware Routing protocols (LAER) and Power Efficient Reliable Routing Protocol for Mobile Ad Hoc Networks (PERRA), is also carried out.

3.2 Node Mobility and Density Classifier

Mobile ad hoc networks have gained a set of interest lately in the investigate community. On the other hand, owing to the different node mobility patterns and density of nodes, such networks have dissimilar connectivity patterns. In conventional MANETs it is implicit that end-to-end paths subsist from any source to any destination most of the time.

In recent times, there has been an endeavor to classify the various types of mobile nodes assuming there is a centralized influence that has whole information of the network and its dynamics. In this work we give a new that classifies density of network and node mobility patterns. This approach provides an attractive insight on the way that mobile nodes operate but it is not practical due to the hypothesis of the centralized mechanism doing the classification.

In this work, we make the primary step to decide the class that a mobile network belongs to in a dispersed style and using only information that the nodes have up to that point in time. Because the mobile nodes frequently have limited energy, computation capability and storage, it is preferable that the number of computations at every one node is tiny. Consequently, our approach is based on an analytical framework which is joint with some simple node observations in order to make a decision the class of the network. Note that in this work we are interested only on the suitable decision of the class that the mobile node belongs to, and it is up to the network designer to describe the way that the nodes could then choose the most suitable routing protocol for this network class.

3.2.1 Classification of Node Mobility and density in MANET

In this section we describe the three states of mobile nodes that we will use throughout the classification process and present the main aspects of Node Mobility and Density Classification algorithm. We will use the results from this algorithm to contrast the accuracy of the proposed model.

Fig. 3.2 Various states of node Mobility

The node mobility classifier set contains the three states as shown in fig. 3.2. They are

? Slow State

? Medium State

? High mobile nodes State

Slow State Mobile nodes in MANET

We use the term Slow State Mobile nodes in MANET to refer to the traditional Mobile Ad hoc networks (MANETs) where it is implicit that the network is associated most of the time due to the mobility of nodes in slow state. This means that at every one timeslot there is an end-to-end path that links every pair of source and destination. In addition, it is unreservedly assumed that the links do not transform that swift, which involves that the routes between sources and destinations do not vary that much. The most general routing protocols that are used in this state are AODV and DSR.

Medium State Mobile nodes in MANET

In the case of the Medium state Mobile Nodes in networks, no contemporaneous end-to-end paths present most of the time and communication is attained by the store, carry, and forward model of routing. Such paths are often said to be space-time paths to distinguish them from the contemporaneous space paths used in MANETs. Many routing models have also been proposed for this class of mobile nodes such as Epidemic Routing and Spray and Wait.

High mobility State nodes in MANET

Such networks are actually sparse and the mobility of the nodes doesn�t permit them to converse even through space-time paths. In fact, the lengths of the space-time paths are too elongated. In this class of mobile nodes it is preferable to use extra mobile nodes that shift around the network area collecting messages and transmitting them to the destination nodes.

Based on the states (i.e, slow state, medium state and high mobility state), non linear programming is applied. Non linear programming is used over the three states and identifies stable link path and minimum energy conserved routes for the transmission.

Density classification

Node density classifier is done by segregating zonal areas based on optimized grid nodal size. Grid nodal threshold is evaluated with heuristic approach by examining regions of ad hoc network for better path stability and minimum energy drain rate of mobile nodes.

3.2.2 Node Mobility and Density Classification Algorithm

In this section we present the Node mobility and density classification algorithm which is based on analytically calculated formulas and simple node observations. First, we present the assumptions and some definitions of quantities that are used throughout the Node Mobility and Density algorithm and then we present the NMDC algorithm.

Throughout the Node Mobility and Density classification algorithm, it is assumed that the mobile network progresses in timeslots. In addition, we suppose that the nodes distinguish in progress the communication range (K). It is indistinguishable for all the nodes and they also identify the size of the network area (N). It will befall obvious later that the algorithm could work with any node mobility model as long as we have analytical formulas for some fundamental elements of the model. We have such formulas in previous works for a broad range of mobility models such as the Random Direction, Random Waypoint, Random Walk and the Community-based model. In this work, without loss of generalization we suppose that the nodes shift according to the Random Direction mobility model.

Let�s now remember how the Random Direction model works. Especially, it moves forward on epochs where at each one epoch every node at first chooses a direction�d�. Next, it selects a speed, v uniformly average speed . Moreover, it selects a duration T of movement from an exponential distribution and it goes towards�d� with the chosen velocity for T time units. If the network frontier is reached, it re-enters from the contrary side of the network. Once T time units it paemploys for a random amount of time selected from [0, Tmax] with average pause time Tstop. After that, a new epoch starts. For expediency, we review the notation used in the Random Direction model as well as throughout the work.

Observe that it is not implicit that the nodes know in progress the number of nodes in the network since as it will be presented, this value could be predictable at every one node by simple observations. Consequently, the proposed NMDC algorithm could adapt to a network in which the number of nodes changes from time to time. This means that the nodes could scuttle the NMDC algorithm from time to time and update their result about the class that the network belongs to, based on the new estimations.

In the following, we demonstrate Node Mobility and density classification algorithm. The derivation of the analytical formulas that is based on LFR is presented in the next section with the analytical framework.

Algorithm 1: Node Mobility and Density Classification

3.3 Lifetime Forecast Routing (LFR)

Lifetime Forecast Routing (LFR) is an on demand source routing protocol that employs battery lifetime Forecast. The objective of this LFR routing protocol is to make bigger the service life of MANET with active topology. This protocol favors the path whose lifetime is maximum. We symbolize our objective function as follow:

---------------> (1)

Where,

: Lifetime of path P

: predicted lifetime of node i in path P

Lifetime Forecast: Each node in MANET tries to approximate its battery lifetime based on its history activity. This is attained using a Simple Moving Average (SMA) predictor by keeping track of the very last N values of remaining energy. The corresponding time instances for the very last N packets relayed by means of every one mobile node. This information is noted and stored in each node. We have suspiciously compared the forecasted lifetimes based on the SMA method to the actual lifetimes for different values of N.

Our motivation in using lifetime Forecast is that mobility brings in different dynamics into the network. The lifetime of a node is a function of remaining energy in the node and energy to broadcast a bit from the node to its neighbors. This metric works well in static networks. Though, it is very hard to efficiently and reliably compute this metric when we have mobility because the location of the nodes and their neighbors continually change.

PSR does not use Forecast and only employs the remaining battery capacity. We believe LFR is superior to PSR since LFR not only captures the remaining (residual) battery capacity but also accounts for the rate of energy discharge. This makes the cost function of LFR more accurate as opposed to just using battery capacity. This is true in MANETs given that mobility could change the traffic patterns through the node, which thus affects the rate of depletion of its battery. Also, recent history in LFR is a good indicator of the traffic through the node and hence LFR chose to employ lifetime Forecast.

LFR is a dynamic distributed load balancing approach that avoids power-congested nodes and chooses paths that are lightly loaded. This helps LFR achieve minimum variance in energy levels of different nodes in the network. As an example, consider the scenario shown in figure 3.3.

Fig 3.3 LFR avoids power-congested paths

Here, node 6 has two flows going through it ( 4?6? , 2?6? and 4?6?). Now, if node 1 wants to transmit data to node 5, the shortest path routing will use 1?6?5. However, LFR will use 1?2?3?4?5 since node 5 is very power-congested (as a result of relaying multiple flows) and the path passing through node 6 will not be selected by LFR.

3.3.1 LFR Route Discovery

In LFR, activity begins with the source node flooding the network with RREQ packets when it has data to send. An intermediate node transmits the RREQ unless:

As a result, intermediate nodes forward only the first received RREQ packet in LFR. The destination node only replies to the initial arrived RREQ because that packet tends to take the shortest path. In LFR, all nodes except the destination compute their forecasted lifetime, Li (Eqn 2) and replace the min lifetime in the header with Li if Li is lower than the existing min lifetime value in the header.

------------------> (2)

where:

: Remaining energy at the ith packet is being sent or relayed through the current node in LFR.

: rate of energy depletion of the current node when the kth packet was sent by LFR and is calculated by as the ratio of the difference between remaining energies of the nodes for packets k-1 and k and the difference between arrival times of these two packets.

N: length of the history used by LFR for calculating the SMA

In LFR model, when an intermediate node receives a RREQ packet, it starts a timer (Tr) and keeps the min. lifetime in the header of that packet as Min-Lifetime (LT). If additional RREQs arrive with same destination and sequence number by LFR, the cost of the newly arrived RREQ packet is contrasted to the Min-Cost. If the new packet has an inferior cost, Min-Cost is changed to this new value and the new RREQ packet is forwarded. Otherwise, the new RREQ packet is dropped by LFR.

Algorithm 2: Functionality in intermediate node

In LFR, the destination waits for a threshold number (Tr) of seconds after the first RREQ packet arrives. During that time, in LFR the destination examines the cost of the route of every arrived RREQ packet. When the timer (Tr) expires, the destination node chooses the route with the minimum cost and replies. Afterward, it will drop any received RREQs. The reply also contains the cost of the chosen path appended to it. Every node that hears this route reply adds this route along with its cost to its route cache table by LFR. Although this scheme LFR-NMDC could somewhat increase the latency of the data transfer, it results in a significant power saving as will be shown later.

LFR has a route invalidation timer that invalidates old routes. This helps in removing elderly routes. This also avoids over convention of particular routes in cases of low mobility.

LFR Route Maintenance

Route maintenance is required for two reasons:

? Node Mobility: Connections between some nodes on the path are misplaced because of their movement,

? Change in forecasted lifetime

In the primary case, a new RREQ is sent out by LFR and the entry in the route cache corresponding to the node that has moved out of range is purged. In LFR, Following policy is adopted to undertake the second situation: Once the route is built, the weakest node in the path (the node with minimum forecasted lifetime at path discover time) monitors the reduce in its battery lifetime. When this remaining lifetime reduces goes beyond a threshold level, the node transmits a route error back to the destination as if the route was provided invalid. The destination sends this route error message to the source by using LFR. This route error message forces the source to initiate route discovery again by using eqn 3. This decision is only dependent on the remaining battery ability of the current node and its liberation rate in the short history. Therefore is a local decision. LFR adopts this local approach because this approach minimizes control traffic. Figure 3 shows an example of route expiration process.

Fig.3.4 a)

Fig.3.4 b)

Figure 3.4: (a) Node 3 sends route error to destination (node 4) (b) send route error to the source (node 1)

More precisely, node i generates a route error at time t when the following condition is met:

Where

--------------> (3)

Li: predicted lifetime of node i

t: current time

t0: route discovery time

d: change threshold for lifetime Forecast

Fig.3.5 Architecture of Lifetime Forecast Routing With Node Mobility and Density Classifier (LFR-NMDC) Model

Fig.3.5 demonstrates Architecture of Lifetime Forecast Routing With Node Mobility and Density Classifier (LFR-NMDC) Model. Lifetime Forecast routing (LFR) with node mobility and Density Classifier Model are designed to stabilize the most of the link and to conserve the energy in the routing protocol. The node mobility classifier set contains the three steps. They are classified as Slow State, Medium State and High mobile nodes State.

Non linear programming set is applied based on the types of node mobility to identify stable link path and to minimize the energy conserved on the routes for the transmission. Node density classifier is done by segregating zonal areas based on optimized grid nodal size. Grid nodal threshold is evaluated with heuristic approach by examining regions of ad hoc network for better path stability and minimum energy drain rate of mobile nodes.

To handle dynamic topology of MANET with ZRP protocol, Lifetime Forecast Routing (LFR) is designed to extend service life of mobile nodes and identify the path with maximal lifetime. Each node estimate its battery lifetime based on its past activities. Simple Touching Average (STA) forecaster keep track of last N values of remaining energy and corresponding time instances for the last N packets received/relayed by each mobile node.

3.4 Performance Evaluation of Lifetime Forecast Routing with Node Mobility and Density Classifier

The performance of proposed Lifetime Forecast Routing with Node Mobility and Density Classifier (LFR-NMDC) Model in a mobile ad hoc network is evaluated in this section.

3.4.1 Simulation setup

We used the event driven simulator ns-2 [3] for experimental evaluations. In this simulation, set up n nodes consistently at randomly surrounded by 1000m � 1000m squares, with n changeable among 100 and 1000 determining the mobile nodes movement patterns. In particular, to exactly estimate the presentation of the system in which each node progress to an randomly selected position with an randomly chosen speed among a predefined minimum and maximum speed.

The moving mobile networks wait there for a specified pause time. After the pause time, it next arbitrarily chooses and travels to another location. This arbitrary progression is constant during the simulation period. All simulations were carried out for 900 simulation seconds, predetermined a pause time of 30 simulation seconds and a minimum moving speed of 5 m/s of each node.

It is considered that all mobile nodes are prepared with IEEE 802.11 network interface card, with data rates of 2 Mbps. Arbitrary connections were created using CBR traffic such that every one node has chance to attach to every other node. Packet size was 512 bytes. The primary battery ability of every node is 100 units.

Simulation parameters taken in the performance evaluation of LFR-NMDC campaigns are listed in Table 3.1.

Table 3.1: Parameters used during simulation

Parameters Value

Area 1000*1000 m

No. of nodes 100-1000

Simulation duration 900 sec

No. of repetition 5 times

Radio transmission range 100 m

Physical/Mac layer IEEE 802.11

Pause time 30 sec

Mobility model Random direction model

Node movement 5 � 35 m/s

Data sending rate 2 Mbps

Each packet 512 bytes

Traffic Type CBR

Table 3.1 lists the general parameters adopted in the simulator in spite of the particular considered protocols. In order to validate the effectiveness of the LFR-NMDC model, some simulations and comparisons with other energy aware protocols have been assessed. In the following, it will be shown how proposed LFR-NMDC model represents a good tradeoff in terms of normalized control overhead, data delivery ratio, network lifetime and average energy consumption in comparison with the other protocols. We vary the mobility and density of nodes and study their effects on these metrics. To take out average values, we simulated every scenario five times.

The performance of the proposed Lifetime Forecast Routing with Node Mobility and Density Classifier (LFR-NMDC) Model is measured in terms of

i) Data Delivery Ratio

ii) Nodes Lifetime

iii) Normalized Control Overhead

iv) Energy Consumption

3.4.2 Simulation Results and Discussions

The performance results of proposed Lifetime Forecast Routing with Node Mobility and Density Classifier (LFR-NMDC) Model is illustrated. A comparison with other known approaches, such as Link-stAbility and Energy aware Routing protocols (LAER) [2] and Power Efficient Reliable Routing Protocol for Mobile Ad Hoc Networks (PERRA) [3], is also carried out.

Node Lifetime

The node lifetime is defined as the time taken for a fixed percentage of the nodes to die due to energy resource exhaustion. In the field of networks, Node Lifetime refers to the unpredicted loss of life in the nodes. In MANET, the nodes lifetime increases when slow mobility state.

Table 3.2 Node Lifetime

Node Mobility (m/s)

Node Lifetime

Proposed LFR-NMDC LAER PERRA

5 98 82 73

10 97 78 71

15 96 80 64

20 96 77 67

25 95 69 61

30 93 63 62

35 92 61 63

The above table (Table 3.2) describes the nodes lifetime obtained when mobility rate increases in the MANET environment. The outcome of the proposed LFR with NMDC model in MANET is compared with an existing Link-stAbility and Energy aware Routing protocols (LAER) [2] and Power Efficient Reliable Routing Protocol for Mobile Ad Hoc Networks (PERRA) [3] for detecting nodes lifetime.

Fig.3.6 Node Lifetime

Fig 3.6 describes the nodes lifetime occurred when mobility rate increases in the Mobile Ad-hoc network. The network lifetime in the proposed LFR with NMDC model is high since we use Lifetime Forecast Routing. LFR balances the energy consumption among the nodes in the network and extend its network lifetime. If the mobility rate becomes low, then the lifetime of nodes from the MANET is high. Compared to an existing Link-stAbility and Energy aware Routing protocols (LAER) [2] and Power Efficient Reliable Routing Protocol for Mobile Ad Hoc Networks (PERRA) [3] for nodes lifetime, the proposed LFR-NMDC model outperforms approximately 21-32% well in MANET.

Normalized Control Overhead

It is calculated as the number of control packets sent in the LFR-NMDC model, LAER and PERRA protocol.

Table 3.3 Normalized Control Overhead

Node Mobility (m/s)

Normalized Control Overhead

Proposed LFR-NMDC LAER PERRA

5 0.4 1.2 2.2

10 0.6 2.5 4.1

15 0.8 2.8 7.4

20 0.9 3.1 11.2

25 1.2 3.2 12.5

30 1.3 3.9 15.3

35 1.7 4.5 16.7

The table 3.3 describes the Normalized control overhead obtained when mobility rate increases in the MANET environment. The outcome of the proposed LFR with NMDC model in MANET is compared with an existing LAER [2] and PERRA [3] for detecting control overhead.

Fig 3.7 Normalized Control Overhead

The curve depicted in Fig. 3.7 testifies the increase in the normalized control overhead for higher speed. It is possible to observe the good scalability of model based on the local topology knowledge such as LFR-NMDC and LAER [2]. The technique applied to both models and the only local control packets exchange (HELLO pkts) determines a similar performance of LFR-NMDC and LAER [2], differently by PERRA [3] that is forced to start new route discovery procedure that increases the control overhead. Proposed LFR-NMDC attains 7-12% less control overhead when compared with LAER [2] and it attains 10-57% less compared to PERRA [3].

Average energy consumption

This parameter allows to make considerations about energy wastage associated with the route maintenance and route discovery and it accounts for energy consumption during transmission and reception of control and data packets.

Table 3.4 Energy Consumption

Stability Weight

Energy Consumption(J)

Proposed LFR-NMDC LAER PERRA

0.1 3.5 5.6 6.3

0.2 3.7 8.3 9.7

0.3 4.2 10.2 11.3

0.4 5.7 12.2 14.8

0.5 6.9 14.7 15.7

0.6 7.5 15.1 17.1

0.7 9.2 17.3 19.3

The table 3.4 describes the Average energy consumption obtained for various stability weights in the MANET environment. The outcome of the proposed LFR with NMDC model in MANET is compared with an existing LAER [2] and PERRA [3] and it shows LFR-NMDC consumes less energy.

Fig 3.8 Energy Consumption

The average energy consumption for different stability weight values are shown in Fig.3.8. It is interesting to observe proposed LFR-NMDC model consumes lower energy: this is due to the simplest topology management and the routing is based on the Lifetime forecast routing. Increasing the stability weights leads to the selection of shorter and more stable routes and this increases the energy consumption of nodes. Proposed LFR-NMDC model improves further the performance reducing the energy consumption about 13-27% in comparison with LAER [2] and about 15-32% in comparison with PERRA [3].

Data delivery ratio

It is the number of packets received at destination on data packets sent by source in LFR-NMDC, LAER [2] and PERRA [3] protocols.

Table 3.5 Delivery Ratio

Number of Nodes

Delivery Ratio (%)

Proposed LFR-NMDC LAER PERRA

100 98.72 86.83 83.02

200 98.2 85.25 81.22

300 97.43 83.92 78.93

400 97.14 82.92 77.73

500 96.68 80.04 77.12

600 95.23 79.47 74.25

700 94.17 76.73 68.24

The table 3.5 describes the data delivery ratio obtained for various number of nodes in the MANET environment. The outcome of the proposed LFR with NMDC model in MANET is compared with an existing LAER [2] and PERRA [3] and it shows LFR-NMDC achieves higher delivery ratio.

Fig. 3.9 Delivery Ratio

Delivery ratio of proposed LFR-NMDC model and existing LAER [2], PERRA [3] protocols for different number of nodes is shown in Fig. 3. Data delivery ratio is the ratio between number of delivered data packets and the number of generated data packets in all nodes. Note that the number of generated packets is the expected number of generated packets. We generate as many as 10,000 data packets during the simulation. They are generated between random source and destination pairs at random times. Many of these might not have reached their intended destination because of lack of existence of a route between the source and destination for various reasons. Also, the network lifetime clearly affects this ratio. If the network was alive for longer time, it implies that more data traffic goes through since we establish random connections throughout the time of the simulation.

Proposed LFR-NMDC model improves further the performance increasing the data delivery ratio about 22-38% in comparison with LAER [2] and about 25-41% in comparison with PERRA [3].

3.5 Summary

A routing protocol called LFR with NMDC, based on the joint metric of link stability and energy consumption, has been proposed. The main objective of LFR-NMDC is to reduce the variance in the remaining energies of all the nodes and thus extend the network lifetime. It attains this by doing local decisions and with minimum control overhead. We demonstrate that LFR-NMDC brings about a clear enlarge in network lifetime.

Its performances have been compared with other two protocols proposed in literature such as LAER and PERRA. LFR with NMDC protocol inherits the scalability, improving the performance in terms of node selection with superior link duration when a higher weight is set to the stability index. LFR with NMDC outperforms LAER and PERRA in terms of control overhead and in terms of a higher capacity to balance traffic load because of the less energy consumption included in the joint metric.