EEM-MAC: Enhanced Energy E�cient Mobility aware MAC protocol for Mobile Internet of Things

: Mobile wireless sensor networks (MWSNs) have become a foremost solution in many emerging applications both in industry and academia. Moreover, considering the mobile node in WSN is a challenging task to designing efficient communication protocols, specifically at a medium access control (MAC) layer. Most of the existing protocols consider only for static and slow mobility. To meet with future MIoT applications, in this paper we propose Enhanced Energy Efficient Mobility aware MAC (EMM-MAC) protocol. Our EMM-MAC protocol consists of 3 contributions i) static synchronization and mobility handling phase to support both environments, ii) By using queue length-based channel access priority for static nodes, and iii) the combined highest signal strength and node status-based channel access priority for mobile nodes without any control packet overhead. The simulation results verify that EMM-MAC yields a notable improvement in the average power consumption, packet latency, and packet


Introduction
The latest development of wireless sensor networks (WSNs) becomes a leading network technology for next-generation smart wireless networks.The global market is dynamically changing and it forecast that connected IoT devices are estimated to 75.44 billion worldwide by 2025 [1] and the global IoT market worldwide is expected to raise 1.6 trillion by 2025 [2,3].This network during the last years coming out of a new paradigm called Mobile Internet of Things (MIoT) consist of a large number of tiny nodes that can Connect both living things and inanimate, sensors for raw data collection and communication over an IP Network through multi-hop fashion.In such MIoT applications, the usage of mobile nodes plays a vital role has broadened their application both in academia, industry, and areas such as wildlife, habitat & patient monitoring, tracking enemy movements, and many other network applications [4][5][6].Under mobile-based applications packet transmission extremely important when connection established, because dynamic nature, changes in topology, and accessing channel for a long time leads to minimizing the sensor battery lifetime.Hence packet transmission to the preferred device (i.e.) sink is more important before losing the connection.To find a suitable solution and reliable communication between nodes Medium Access Control (MAC) layer plays a vital role, this makes a challenging role to enhance MAC under mobile environment for future mobile WSN.
Over a year many WSN based MAC protocols have been designed, such as S-MAC [7], T-MAC [8], and B-MAC [9], and provide solutions for low power energy consumption and duty cycle scheduling mechanism, etc., under static environment, where a topology remains fixed meanwhile depending on the physical medium may change (i.e) energy hole problem.Previous research works have been addressed towards static to the mobile environment of which some of the prominent reviews are presented in [10-13] [7-10].The survey in [10] [12] presents various mobility models, estimation, and detection techniques concerning the challenges in the design of mobility MAC protocols in MWSNs.In [11] presents key aspects of mobility support at the MAC, network layers.Under mobility, the node's position changes dynamically may impact high processor computing, unwanted radio wake up and decrease in network lifetime.To overcome this, they come with a proposal called Network of Proxies (NoP), to competent of managing low processing, energy, and less timeconsuming.In [13] mobility MAC protocols are classified into two groups namely synchronous and asynchronous schemes in MWSNs.Based on this the major challenges and protocol design for future consideration they addressees.However, to the best of our knowledge, very few observe the vital role in mobile IoT applications of dynamic nature.
Mobility MAC protocol still endures three problems.The first problem is a classification of stationary node (SN) and mobile node (MN) and secondly under routes reconstruction may cause packet loss (i.e., remaining (n-1) packets that a node contains in it transmit buffer) when MN moves from the virtual region (VR) to other.Lastly, duty-cycle medium access control (MAC) protocols allow the neighboring nodes to be synchronized effectively under static and struggle to handle under dynamic.In this paper, we design an enhanced Mobility Based queue aware Mac protocol called EEM-MAC to reduce the above state problems.To achieve EEM-MAC focus on low power short and long preamble based on queue length MAC mainly due to the dynamic's nature induced by mobile nodes in a network.
Consequently, the static node can provide known information about the radio (e.g., buffer length, power, and Radio Signal Strength Indication) module.By using this information, we adopt a short preamble-based duty cycle mechanism for quick and efficient data transmission under normal mode.We further introduce a long preamble-based mechanism under the mobile node that can be stimulated to allow seamless handover to the neighboring node to make the radio receive packets till the ongoing data transmission on the respective channel in a network.
Similar to the existing X-MAC [14] EEM-MAC is an asynchronous MAC protocol, data transmission from sender to receiver based on buffer length.When a sender attempts to transmit the packet to the receiver, once the sender receives the link-layer acknowledgment it broadcast the queue length to neighboring nodes in the respective network region.Hence by knowing wake-up time allows the neighboring node to adjust their radio goes to sleep state.Hence EEM-MAC increases its sleep time to save energy by reducing the unwanted wake-up time (i.e.) ideal mode.
The contributions presented in this paper are threefold:  After a detailed study of existing mobility-aware MAC protocols for WSNs, we propose an asynchronous duty cycling mobility MAC protocol, called EEM-MAC, which adopts a two-phase: data transfer and mobility handling.To achieve this, we embedded S-static and M-mobility flags (2-bit) in the reserved field in IEEE 802.15.4 MAC header [15]. In the data transfer phase, a sender transmits data to the receiver based on queue length information and the respective sender will broadcast to use queue length of a neighbor node to adjust the data transmission to go to a sleep state until the ongoing transmission on the channel.Therefore, EEM-MAC behaves like static MAC but improves the energy efficiency and increases the network lifetime. In the mobility handling phase, node movement is realized by using obtainable information from the radio module (e.g., RSSI, position).By predicting the node mobility, it will set mobility flag, if the flag is perceived, the mobile node will choose the preferred node and extent the long preamble time.By adopting a long preamble time receiver or potential next-hop node will remain inactive state and the non-targeted node will not affect the channel and go back to sleep mode and.Hence EEM-MAC provides fast discovery, good packet reception, and minimum delay under mobility and static environment.
 Finally, to determine the performance of EEM-MAC, we perform our proposed protocol in Contiki OS (top of COOJA) on well-known WSN platforms (Z mote).In addition, we evaluate our protocol design comparing it with existing mobility MAC protocols in two different mobility models and scenarios.
The remainder of this paper is structured as follows: we provide a detailed summary of the most pertinent related works and problem formulation in section 2. In section 3, we present the system model and detailed analysis and design of the EEM-MAC protocol.Section 4 describes the performance evaluation of the EEM-MAC protocol against well-known MAC on top of the Contiki OS.Finally, section 5 provides conclusions and potential suggestions for future work.

Related works:
During the previous decade, quality in WSNs was improved of which MAC protocols play a vital role because they directly access the medium.Many solutions were designed related to mobility-aware MAC in MWSNs.However, they are categorized into scheduling, preamble duty cycle, synchronous, asynchronous, and hybrid.Hereafter, our studies focus on recent and relevant approaches towards mobility-aware preamble-based MAC protocols for MWSN to our investigation.
In [16] author proposed mobility aware medium access control protocol called MA-MAC that is extended from X-MAC [14].To support mobility MA-MAC defines two thresholds: the first threshold is to enable the handover initialization and the second threshold is used to activate the intermediate node (next hop distance) and inform to receive data from the mobile node.Under static mode data transmission is based on a preamble-based mechanism, when the node completes the data it transmits ACK once receive goes to sleep state.On other hand during mobility, the sender will embed the handover bit and broadcast during ongoing packet transmission.During this first potential receiver will receive the bit and the mobile node transmit the data to the region.After a random interval remaining receiver will wake up and receive the handover bit followed by the first.Thus the mobile node can perform uninterrupted transmission in a structured mechanism.
Limitations of the protocol: a) when a mobile node broadcasts the handover bit during mobility the first potential receiver will receive data and if the adjacent/next hop receiver fails to wake up makes the mobile node decrease in throughput.b) Author highlights preamble cycle it holds perfect for static mode by adopting the random wakeup interval time at the same time under mobility mode random wakeup interval makes non targeted node/ targeted node will be in between active/ideal state leads to more energy consumption and decrease in network lifetime.
In MT-MAC [17] is an extended version of T-MAC [8].MT-MAC resolves high packet drop and can balance tradeoffs between energy, latency.To solve the connection problem with the mobile node they use RSSI and LQI (Link Quality Indicator) values in the SYNC packet.
During mobility when MN moves to a new region and respective MN must wait for t time to choose sink in the respective region, this makes low throughput and high packet drop.Thus MT-MAC adopts 1. Scheduling phase 2. Mobility handling phase.Before start transmission, it adopts three different identifier flag styles namely border node, cluster head, stationary node, and the mobile node.At the scheduling phase, nodes start to exchange the SYNC control packets.After exchanging SYNC packets, they are categorized into three types: first when the node does not receive SYNC packet from a neighbor in the network region and it will independently schedule a frame and broadcast intern act as cluster head node type.Secondly, when the node adopts the frame schedule with the neighboring node it sets node type as a stationary node.Finally, if the node has a different frame schedule it sets node type as border node (BN).Hence data transmission under static (SN flag is set) environment is scheduled within the virtual region (VR) and if the node travels from one VR to another it identifies based on the RSSI and LQI threshold values.once the threshold varies the respective flag from SN changes to mobility phase (MN flag is set).To achieve smooth handover MT-MAC uses a handover bit along with a SYNC packet.Once it broadcast neighboring nodes and border node adopt a new schedule for smooth communication without delay.
Limitations of the protocol: a) Data transmission based on identifier flags (overhead) and are suitable for the targeted node or sink if the non-targeted node pursues the same schedule makes the node in the wake-up state leads to the dead node.Hence MT-MAC is well-matched for non-energy-constrained applications.b) Under mobility, the interaction between MN and BN makes BNs node consume more energy and further unwanted wakeup (active and ideal state) between neighboring nodes build packet loss due to its less network lifetime.
In [18], the authors present the MS-SMAC protocol which is inherited from S-MAC [7].The probability of network disconnection and packet loss occurs during mobility for both mobile and neighboring nodes.To overcome this MS-SMAC adopt MSYNC and MACK in the SYNC packet.MSYNC examine the speed of the mobile node in the respective communication range and broadcast to the receiver.On another hand, the MACK control packet involves 1) determines when to transmit the MSYNC packet.2) Two sleep Times (1and 2).SleepTime1 schedule for sender node in respective cluster region similarly sleepTime2 schedule for new cluster region under mobility.Hence mobility prediction is calculated based on RSSI values of mobile nodes in a cluster.When the mobile node enters from one region to another it will inform its position, next to sleep time, next interval, and speed.The border node will receive this up-to-date information so that the mobile node will ensure better connectivity.
Limitations of the protocol: a) overhead involves computation towards sleep time, interval time, node speed, and position.b) Control packet (MSYNC and MACK) is flooded innetwork particularly for mobile node but in some situation, if more mobile nodes in network the synchronization between static, neighboring, and border node may fail due to complex computation.This leads to high packet drop and high energy consumption due to an unwanted wakeup state.
In [19] Ba et al. present the MOX-MAC as an extension of the X-MAC protocol [14] which uses a preamble-based data transfer mechanism for both static and mobility conduction.
Under the static condition when a static node is attempting to transmit data it checks the channel is free or not.Once it receives the preamble it will communicate to the respective node with backoff time.Similarly, under mobility, when the mobile node is attempting to transmit the data it will sense the medium to receive the ACK control packet from the respective static node.
But mobile node will wait for some time till the ongoing transmission from the respective static node, once the node becomes free subsequently it sends the data packet to the receiver static node.One major problem is that when the static node has long ongoing transmission in such case the mobile node must keep their radio ON and leads to makes unnecessary energy consumption and high packet drop.Fig. 1 shows when the mobile node wants to transfer the data it waits for the random interval to make the connection establishment to receiver static/ sink node.In such cases when the collision or channel access to some other nodes make more packet loss and high energy consumption.Similarly, if the ACK also fails leads to unwanted wakeup and introduces a long delay in a network. It was suitable only for selected applications.

challenges
In WSNs either in designing protocols or applications they focus on static networks, The ultimate intent of the energy model is to accurately measure the battery power consumption of the sensor node during the different operating states.A sensor node comprises a transceiver module, microcontroller, a memory element, analog-to-digital converter (ADC), battery, and sensing units.Among the sensor sub-components, the transceiver unit consumes a significant source of battery energy due to the transmission and reception of data packets between sensor nodes.The total energy required by the transmitter ( ) to forward -bits of data over a distance ′' is expressed as Where  () is the energy consumed by the microcontroller unit and  (, ) is the energy consumed by the amplifier to transmit -bits over a distance ′′.Equation ( 1) is modified and rewritten concerning propagation coefficients as follows Where  and  are free space and two ray propagation coefficients, ′ is the distance between transmitter and receiver nodes, and ′′ is the reference distance.Total energy consumed during the reception ( ) of  bits of data packets can be represented as Where  is the energy consumed by the sensor node to receive packets and ERDC is energy dissipated during the data acquisition and compression process.Total energy dissipated  by the sensor node during the idle listening time can be represented as Where  is the idle waiting period,  _ is the energy consumed by the transceiver module to remain in the idle wake-up mode, and  is the energy consumed by the memory element to hold the packets.It is widely proven [23] that the energy consumption during  is equivalent to that of  or  process.Thus, minimizing  is crucial to improve the battery efficiency of the sensor node.

Mobility Model
In this section, the mobility model describes the pattern of the mobile node and predicting their localization, direction, and speed concerning time.Since the mobility model play an essential role to determine the significant performance of mobility MAC in MWSN [24].The main objective to study mobility patterns is used to relate to real-life applications realistically.Consider two nodes ( ,  ) situated at ( ,  ) and ( ,  ) that  ∈ ( ,  );  ( ,  ).The node p and q move in a particular direction with variable rate making an angle   i.e. node  moves a distance 1 and node  moves a distance d2 in interval T = 0 to t.At initial time t, the nodes located at a distance  ( ,  ) and  ( ,  ) is given as, Let the node  and  moves with the variable rate  and  making an angle  and  , the node distance  and  in a particular time t is given as,  =  ×  and  =  × After some time,  = , mobility of the node moves to a new position and the updated position reach by node  is given as

𝑣 , = 𝑢 , + 𝑣 × 𝑡 × 𝐶𝑜𝑠𝜙
Similarly, after some time  = , mobility of the node moves to a new position and the updated position reach by node  is given as

𝑣 , = 𝑢 , + 𝑣 × 𝑡 × 𝐶𝑜𝑠𝜙
Once the mobile node reaches the new position, the updated distance between the nodes  and  is given as

Dynamic Agile Mobile Node (DAMN) Prediction with EEM-MAC
In general IEEE 802.15.4 standard has 2 consecutive CCA checks in an interval along with the back-off period.The main role for consecutive CCA checks is for signal detection in the wireless medium [25].Generally, sensor mote will wake up their transceiver radio module from its sleep state when double CCA checks operation.Once the radio wakes up it checks the channel availability, if the channel is free it will start data transmission towards the sink by placing the radio module inactive state.The obtained value (channel condition, RSSI) received at the beginning is applicable only for the ideal and no interference conditions.Next, if interference occurs the RSSI and channel condition may not be valid and leads to misdirection for neighbouring nodes in a dynamic network.At last, CCA struggles to sustain its precision when congestion and false wakeup by nodes.To predict effectively under dynamic nature, we introduce DAMN mode which further reduces the false prediction, false scheduling mechanism between sender and receiver node in a network.Initially before starts sending data, it waits for  seconds to collect the node information (RSSI, queue length, and position).

Simulation results and discussion
In this section, we discuss and evaluate the performance of our proposed enhanced mobility aware EEM-MAC protocol as compared it with well-known protocols, i.e., M-ContikiMac [21], ME-ContikiMac [22], and MoX-MAC [19].To validate, we use ContikiOS End to End delay is the time taken by the data packet to reach from source to sink node with acknowledgment.Energy consumption is equal to per-packet energy dissipation of sensor node when a microcontroller is an active mode as described in section 3.1.In ContikiOS, an inbuilt power trace module is available to calculate per packet energy consumption of the respective mote.Generally, when a radio module is in an OFF state, the sensor is in low-power mode (LPM) mode and the energy consumption is 20µA.
when the radio is switching from OFF to ON state it is considered to be in CPU/Ideal state.Meanwhile, energy consumption also depends on transmission current (Tx) and reception current (Rx).The formula to calculate energy consumption for each packet as follows: By considering Table 3.The consumed energest value of node is calculated in simulation models by equation (13).

Figure. 1
Figure. 1 MoX-MAC data transmission between the static and mobile node Papadopoulos, G. Z et al. presented the M-ContikiMac [21] which is based on ContikiMac [20] which is a default MAC protocol in ContikOS [26].ContikiMac is an asynchronous preamble-based protocol designed for static domains.For mobility, they adopt the burst mode transmission in a network.When a node decides to transmit data first it informs the receiver by setting the flag (UniByte).On the monitor, this flag the receiver make the node keep their radio in ON state to receive the n packets from the mobile node until the ongoing transmission.Fig 2. Shows the M-ContikiMAC procedure, initially, the sender S wants to transmit data under burst transmission.However, to transmit data sender must discover the potential static node in its network region before moving to a new region.Thus it adopts anycast transmission method so that it places the destination address in the packet.Initially, the first packet incorporates an additional UniByte flag (flag=0) and it continuously transmits to find the static neighboring nodes.Meanwhile, R1 checks the flag makes the radio wake up, and starts receiving the packets of static nature.Under Mobility, the mobile node receives ACK it enables the UniByte flag (flag=1) and transmits the remaining packets to R1 before

Figure. 2
Figure. 2 M-ContikiMAC procedure for the sender and mobile node Set back in M-ContikiMac protocol leads to design ME-ContikiMac [22].To avoid packet duplication at the mobile node it adopts the control packet.In ME-ContikiMac when a mobile node begins the transmission it does not accept the ACK packet but it will adjourn the ongoing preamble cycle according to n packets.First data transfer is the same as a previous mechanism [21] under static nature, on the other hand when two static node S2, S3 sample the medium and receive the control packet the ACK and transmit under collision when the on two static nodes S2 and S3 receive the control packet from M at the same time ACK packet

Figure. 3
Figure. 3 ME-ContikiMAC procedure for the sender and mobile node Table. 1 Comparison of different mobility aware protocols concerning mode (Ssynchronous, A-Asynchronous), derived from, throughput, latency, mobility detection, and comments Figure. 3 (a) Mobile sensor node movement through different VRs (b) Mobile handover process in VR


At this time DAMN further extends the Channel Check Rate (CCR) to get better RSSI threshold values set over the noise floor to detect sender signals.The subsequent steps are as follows: Initialize the NETSTACK_RADIO once wake up and start sampling to channel to get the average RSSI value Assign the RSSI threshold value to as previous obtained RSSI. Increase the transmission radio power and assign the threshold value to the obtained signal level energy  Again sample the channel and update value based on a recent CCR interval check.The detailed flow diagram of DAMN mode is shown in Fig 4.

Figure. 4
Figure. 4 Flow diagram of the DAMN mode3.4.Static synchronization phase with EEM-MACThe proposed EEM-MAC implements a channel access priority mechanism combined with a receiver coordinated radio sleep scheduling approach among the competing sensor nodes under each VR.The EEM-MAC implements distinct channel access for static and mobile sensor nodes.In the static node scenario, the receiver node 'R' collects the queue length

Figure. 5 10 )3. 5 .
Figure. 5 Sender and receiver data transfer in static condition As shown in Fig. 5, sensor nodes 'S1', 'S2', and 'S3' advertise their queue length with the receiver 'R' during the static synchronization process.The receiver 'R' initiates the access priority to the node 'S2' with a short preamble based on the packet count in the buffer and broadcast sleep message among the competing neighboring nodes 'S1' and 'S3'.The neighbouring nodes remain in the sleep state until the data transmission between 'S2' and 'R'.Let TSD is the time interval for successful packet transmission, QL is the queue length information of the current node, Tack is the successful acknowledgment duration and  ,

Figure. 6
Figure.6 IEEE 802.15.4 MAC header with MN field The receiver node 'R' initiates access priority to a mobile sensor node with the highest RSS value and adopts a long preamble mechanism to ensure high data transfer with minimum loss probability during the node movement between the VRs.When the RSS value of the existing mobile sensor node falls below the minimum sensitivity value, the receiver node 'R' dissociates with the node.Conversely, the competing neighbor nodes remain in the sleep state

Figure. 7
Figure.7 Sender and receiver data transfer under node mobility conditions As shown in Fig. 7, sensor nodes 'S1', 'S2', and 'S3' advertise their mobility state and queue length information during the synchronization process.The receiver 'R' initiates the access priority to the mobile node 'S2' and adopts a long preamble mechanism for elongated data transfer during node movement.The mobile node long preamble duration  , is represented as  , = ( ( ) +  ) (11) Meanwhile, the receiver 'R' broadcasts sleep messages to the competing 'Si-1' neighbour nodes 'S1' and 'S3' that allow them to remain in the sleep state until the preamble period.The controlled sleep by the competing 'Si-1' nodes substantially minimizes the idle radio listening time, and channel contention that results in minimum packet latency and reduced battery power consumption.By considering the following description, the detailed procedure of EEM-MAC protocol is presented in Algorithm I. R → Receiver

[ 26 ]Figure. 8
Figure. 8 MWSN of the grid and random scenarios (both static and mobile sensors)  Packet delivery ratio (PDR) is calculated as the total packets received successfully by the receiver node to the packets received by the sender.

Figure 9 Figure. 9 Figure. 10 Figure. 11 Figure. 12
Figure9shows the impact of speed in grid topologies under the RWP mobility model.Here results were obtained for MoX-MAC, M-ContikiMac, ME-ContikiMac, and EEM-MAC by varying between normal to high speed ((i.e.) 2m/s to 15m/s).Fig9ashows an impact of varying speed on PDR over grid topologies under the RWP model respectively The average PDR in grid topology was found to be 23.87%improvement for the EEM-MAC as compared to MoX-MAC, 10.24% improvement for the EEM-MAC as compared to M-ContikiMac and 1.69 % improvement for the EEM-MAC as compared to ME-ContikiMac.Figure9bshows the end-to-end delay and it observed that EEM-MAC performs better and reaches the sink faster by an average of 31.9% as compared to MoX-MAC and an average of 25.89% as compared to M-ContikiMac and 2.47% as compared to ME-ContikiMac by varying speed of the MNs.As shown in Fig.9c, it is found that proposed EEM-MAC consumes on an average 54.44% less power as compared to MoX-MAC, on an average 15.88% less power as compared to M-ContikiMac and on an average 6.47% less power as compared to M-ContikiMac in the grid topology.

Table . 3
CC2420 Radio Current Consumption parameters