Optimizing the MAC Protocol in Localization Systems Based on IEEE Networks

Transcription

1 sensors Article Optimizing MAC Protocol in Localization Systems Based on IEEE Networks Juan J. Pérez-Solano *, Jose M. Claver Santiago Ezpeleta Departament d Informàtica, Universitat de València, Avd. de la Universitat, Burjassot, Spain; jclaver@uv.es (J.M.C.); Santiago.Ezpeleta@uv.es (S.E.) * Correspondence: juan.j.perez@uv.es; Tel.: ; Fax: Received: 2 June 2017; Accepted: 3 July 2017; Published: 6 July 2017 Abstract: Radio frequency signals are commonly used in development indoor localization systems. The infrastructure se systems includes some beacons placed at known positions that exchange radio with users to be located. When system is implemented using wireless sensor networks, wireless transceivers integrated in network motes are usually based on IEEE stard. But, CSMA-CA, which is basis for medium access protocols in this category communication systems, is not suitable when several users want to exchange bursts radio with same beacon to acquire radio signal strength indicator (RSSI) needed in location process. Therefore, new protocols are necessary to avoid packet collisions that appear when multiple users try to communicate with same beacons. On or h, RSSI sampling process should be carried out very quickly because some systems cannot tolerate a large in location process. This is even more important when RSSI sampling process includes measures with different signal power levels or frequency channels. The principal objective this work is to speed up RSSI sampling process in indoor localization systems. To achieve this objective, main contribution is proposal a new MAC protocol that eliminates medium access contention periods decreases number packet collisions to accelerate RSSI collection process. Moreover, protocol increases overall network throughput taking advantage frequency channel diversity. The presented results show suitability this protocol for reducing RSSI garing increasing network throughput in simulated real environments. Keywords: fingerprinting; localization; MAC; multi-channel; wireless sensor networks 1. Introduction Nowadays, development accurate indoor location systems is great importance in realization a wide range applications, such as: indoor navigation systems, location aware services, collaborative autonomous robots, etc. The implementation such systems using radio frequency signals is a challenge due to physical properties se signals it has been treated from different point views. As a result, a large quantity research papers proposing distinct methods algorithms can be found in bibliography [1 4]. This work is focused on fingerprinting localization systems [5,6]. A main feature this group algorithms is initial collection RSSI acquired at specific positions in covered area. The RSSI measures are taken using a mobile radio transceiver that exchanges data with infrastructure beacons. Once database with all se measures is conformed, new users can locate mselves in this scenario comparing ir own RSSI samples with RSSI in database. However, continuous tracking users requires exchange a burst multiple data to gar RSSI at every new step. Moreover, as it is shown in references [7,8], achieved location accuracy can be improved if RSSI Sensors 2017, 17, 1582; doi: /s

2 Sensors 2017, 17, are taken at both motes, (i.e., beacon user), with different signal power levels frequency channels. The total number transmitted at every new position in this last case increases drastically. In addition, it represents a challenge because several users may be trying to access medium to exchange with same or with or beacons at same time in same broadcast domain. Wireless sensor networks (WSN) is a key technology in development location tracking systems. WSN motes are simple low cost devices capable performing wireless communications. Usually, wireless transceivers integrated in WSN motes are IEEE [9] compliant. This stard was conceived to support design low-rate personal area networks. In this kind networks contention-based protocols are usually implemented at medium access control (MAC) layer. In this context, well-known carrier sense multiple access with collision avoidance (CSMA-CA) algorithm, proposed in IEEE stard, is most common MAC protocol used in low-rate WSN. Neverless, contention-based protocols present serious restrictions in terms maximum bwidth network throughput when number nodes transmission rate grows due to increasing number packet collisions [10 12]. On or h, time division multiple access (TDMA) MAC protocols can be considered as an alternative to implement MAC layer. These protocols are more suitable for high-rate applications because medium access is divided in time slots that are assigned to nodes individually, avoiding contention periods packet collisions. However, main drawback TDMA [13] protocols is tight clock synchronization that y require to identify starting points every time slot. The implementation such time synchronization protocols in WSN is a great challenge due to low resources capabilities network nodes time uncertainty associated to packet transmission s. Anor approach that can be taken into account to increase overall channel bwidth is based on use different frequency channels. In this sense, reference [14] presents a protocol specifically designed for WSNs that assigns a frequency channel a time slot to every mote, increasing significantly network throughput. However, this protocol presents long s in dense networks. In [15] authors propose a channel-hopping technique combined with contention-based access for packet transmissions. Noneless, estimation a suitable length for schedule is not an easy matter this fact affects delivery latency. The work presented in [16] takes advantage different frequency channels to provide a conflict-free schedule, but channel assignment transmission scheduling needs complex algorithms, especially in dense networks. In [17] a tree network is split in separate smaller networks that make use dedicated frequency channels, but network topology does not consider mobile nodes. The protocol presented in [18] builds a tree it allows that nodes at same level in tree compete to transmit within a shared slot. In [19] a centralized MAC protocol specifically designed for implementation indoor location systems is proposed. In this protocol, a special central node orchestrates transmission beacons to mobile nodes using a special trigger message. The main features this protocol are: (a) medium access is based on TDMA, assigning different slots to each node, (b) are transmitted using only one frequency channel, (c) network topology is restricted to case where all nodes are on same broadcast domain, which limits use this approach on wide areas. Moreover, use TDMA requires strict time synchronization among different nodes. As a conclusion, it is observed that all se protocols do not meet localization system requirements, since y deal with static or specific topologies y introduce medium access protocols with tight time synchronization constraints, which in general are too complex for our purposes. In this work, a first approach to implement underlying location system infrastructure is based on CSMA-CA protocol, as it is presented in IEEE stard [9]. However, due to low performance that this approach provides, a new MAC protocol is proposed. This new protocol is adapted to localization system requirements its main objective is reduction overall that appears during RSSI sampling process. To achieve this end, protocol has to cope with changes in network topology transmissions at different frequency channels. But it also has to

3 Sensors 2017, 17, provide a low computational cost, because it has to be implemented in motes with limited resources. The proposed protocol has been evaluated simulating different numbers users transmitted during RSSI sampling process. Sensors 2017, 17, The rest this paper is organized as follows. Section 2 is devoted to presenting an extensive analysis limited MAC resources. protocol The proposed protocol in has IEEE been evaluated simulating stard. different Section numbers 3 provides users a new MAC protocol specifically transmitted tailored to during implementation RSSI sampling process. RSSI localization systems based on IEEE The rest this paper is organized as follows. Section 2 is devoted to presenting an extensive networks. Results analysis showing MAC protocol performance proposed in improvement IEEE instard. terms Section network 3 provides throughput a new data collection MAC in protocol simulated specifically tailored real environments to implementation are presented RSSI localization in Section systems based 4. Finally, on IEEE concluding remarks are given in networks. SectionResults 5. showing performance improvement in terms network throughput data collection in simulated real environments are presented in Section 4. Finally, concluding remarks are given in Section IEEE MAC Analysis 2. IEEE MAC Analysis Considering unslotted version IEEE stard, medium access algorithm Considering unslotted version IEEE stard, medium access algorithm follows flowchart shown in Figure 1. follows flowchart shown in Figure 1. Figure 1. Unslotted version IEEE CSMA-CA protocol. Figure 1. Unslotted version IEEE CSMA-CA protocol. The parameters involved in this algorithm are: NB: number times a mote has tried to access channel. BE: The backf exponent represents number backf periods mote must wait before accessing channel. The minimum (aminbe) maximum (amaxbe) in stard are 3 5, respectively. macmaxcsmabackfs: Maximum number channel access tries. This value is 4 by default. The medium access process starts determining initial backf period that mote has to wait before checking channel state. This number is a rom value in interval (0 to 2 BE 1). Initially, BE is 3 backf period would be in range from 0 to 7. Next, physical layer performs a clear channel assessment (CCA). If channel is free, mote can start packet transmission, orwise NB BE value are increased by one. The new BE value in this last case is calculated as BE = min (3 + 1, 5), because it cannot be greater than 5. On or h, NB must be always lower than macmaxcsmabackfs. If NB exceeds this limit, transmission is cancelled protocol informs upper communications layers about this fact. The parameters involved in this algorithm are: NB: number times a mote has tried to access channel. BE: The backf exponent represents number backf periods mote must wait before accessing channel. The minimum (aminbe) maximum (amaxbe) in stard are 3 5, respectively. macmaxcsmabackfs: Maximum number channel access tries. This value is 4 by default. The medium access process starts determining initial backf period that mote has to wait before checking channel state. This number is a rom value in interval (0 to 2 BE 1). Initially, BE is 3 backf period would be in range from 0 to 7. Next, physical layer performs a clear channel assessment (CCA). If channel is free, mote can start packet transmission, orwise NB BE value are increased by one. The new BE value in this last case is calculated as BE = min (3 + 1, 5), because it cannot be greater than 5. On or h, NB must be always lower than macmaxcsmabackfs. If NB exceeds this limit, transmission is cancelled protocol informs upper communications layers about this fact. Anor important period time that must be considered during a packet transmission is inter frame space (IFS). IFS are related to packet size in IEEE So, re is a short IFS

4 Sensors 2017, 17, (SIFS), which is applied if packet is short its length is lower that amaxsifsframesize (18 bytes). In contrast, IFS becomes long IFS (LIFS) when packet length is higher than amaxsifsframesize Transmission Delay In packet transmission, seven different terms can be distinguished. Each one sts for a different transmission stage during overall packet delivery process. In this way, total can be divided in following terms: Delay = T Backf + T CCA + T TA + T packet + T TA + T ACK + T IFS (1) where se terms st for: (a) T Backf is back f period; (b) T CCA is time to perform CCA, which it is usually 128 µs in Telosb motes [20]; (c) T packet is packet transmission time; (d) T TA is turn around time that allows device to switch from transmit mode to receive mode vice versa, this time is normally 192 µs in Telosb motes; (e) T ACK is time for transmission an ACK packet; (f) T IFS is final that can be equal to SIFS or LISF times depending on packet length. In case having a transmission that does not require an acknowledgement, T TA T ACK disappear in (1). The most significant term in previous list s is T Backf, since it cannot be known a priori due to its rom nature, because it is usually greater than rest terms. The backf period can be formulated as: T Backf = BO slots T BOslots (2) where first term BO slots is rom number backf slots, which mote includes before packet transmission, T BOslots is duration one slot. The number BO slots is a rom value uniformly selected in interval (0 to 2 BE 1) initial mean value this term is 3.5. In addition to all s involved in a packet transmission, re is anor contribution coming from operating system (OS). Thus, when a packet is sent (received), OS must process packet forward (receive) information to (from) final application. Therefore, two additional s appear at beginning end expressed in (1). These s are: (a) send time at transmitter required to assemble message give transmission order to MAC layer (b) reception time at receiver to process message forward this information to destination application Performance Simulation Localization algorithms require exchange many data with infrastructure beacons to get necessary RSSI. In this context, assessment IEEE networks performance is important to know maximum bwidth packet exchange rate that can be achieved. In this experiment, maximum network throughput is determined using Cooja [21] simulator TinyOS [22] components running on Telosb motes. The network topology considered is a star, where receiver all senders are in same broadcast domain packet length is fixed equal to 23 bytes. The senders are continuously sending to receiver at a constant transmission rate. Since all motes share same channel y are in same broadcast domain, collisions may occur. The CSMA implementation for CC2420 transceiver in TinyOS distribution differs slightly from stard version. In this case, backf slots are selected using a rom number (r) calculating: BO slots = r mod (31 CC2420_BACKOFF_PERIOD) + CC2420_MIN_BACKOFF (3) By default, constants CC2420_BACKOFF_PERIOD CC2420_MIN_BACKOFF are both equal to 10. Since T BOslots is measured with a 32 KHz timer, T Backf is in range from 0.32 to 10 ms. If channel is busy, back-f changes it is computed as:

5 Sensors 2017, 17, BO slots = r mod (7 CC2420_BACKOFF_PERIOD) + CC2420_MIN_BACKOFF (4) Sensors 2017, 17, With that are in interval from to 2.5 ms. In se conditions, transmission follows unslotted BOslots CSMA = r algorithm mod (7 CC2420_BACKOFF_PERIOD) depicted in Figure 1. + The CC2420_MIN_BACKOFF evaluation conducted estimates (4) aggregate network With throughput that are defined interval as from percentage to 2.5 ms. time In se in which conditions, channel transmission occupied with a successful follows packet transmission. unslotted CSMA algorithm Figure 2depicted shows in Figure obtained 1. The results evaluation with conducted three different estimates numbers aggregate network throughput defined as percentage time in which channel is occupied senders packet transmission rates. As it can be seen, initially when number senders increases, with a successful packet transmission. Figure 2 shows obtained results with three different network numbers throughput senders also rises, packet especially transmission for rates. lowas transmission it can be seen, initially rates. when However, number throughput decreases senders when increases, transmission network ratethroughput increases also too much rises, especially due to packet for low collisions, transmission especially rates. when number However, senders throughput is high. In decreases any case, when it should transmission be noticed rate increases that none too much due simulated to packet scenarios collisions, especially when number senders is high. In any case, it should be noticed that none exceeds 25% network throughput. As a result, it can be concluded that basic CSMA simulated scenarios exceeds 25% network throughput. As a result, it can be protocol isconcluded not suitable that for applications basic CSMA that protocol require is not high suitable transmission for applications rates with that multiple require high users trying to communicate transmission withrates with same multiple receiver. users trying to communicate with same receiver. Figure 2. Aggregate network throughput for different transmission rates number nodes (U). Figure 2. Aggregate network throughput for different transmission rates number nodes (U). The aggregate network throughput is defined as percentage time in which channel is The aggregate occupied network with a throughput successful packet is defined transmission. as percentage time in which channel is occupied with a successful packet transmission. 3. New MAC protocol 3. New MAC 3.1. Initial protocol Assumptions it was stated in previous section, stard CSMA protocol is not appropriate in 3.1. Initial Assumptions localization systems with many users beacons, where users have to acquire RSSI at both As it sides was(beacon stated in user mote), previous with multiple section, power levels stard different CSMAchannels. protocol Consequently, is not appropriate in new proposed MAC protocol has been conceived to speed up this process eliminating contention localization systems with many users beacons, where users have to acquire RSSI at both periods in channel access. As it has been mentioned, collection RSSI between a sides (beacon user user a beacon mote), comprises with a multiple sequence power levels exchanges different using several channels. frequency Consequently, channels new proposed MAC power protocol levels. In this hascontext, been conceived a exchange to speedis up defined this as process consecutive eliminating transmission contention two periods in channel access. in opposite As directions has been(one mentioned, from user to beacon collection anor from RSSIbeacon to between user). In this a user a way, this sequence exchanges at different channels power levels represents a burst beacon comprises a sequence exchanges using several frequency channels power levels. transmission, since it includes a continuous sequence multiple exchanges that occupies In this context, medium a for a certain exchange period is defined time. Next, as proposal consecutive this transmission new access protocol two is carried out in opposite directions following (one from user next assumptions: to beacon anor from beacon to user). In this way, this sequence exchanges There is ata different channel, denoted channels Ch1, which power is used for levels signaling represents purposes. a burst transmission, since it includes a continuous Beacons announce sequence ir availability multipleusing exchanges broadcast that. occupies medium for a certain period time. Next, proposal this new access protocol is carried out following next assumptions: There is a channel, denoted Ch1, which is used for signaling purposes. Beacons announce ir availability using broadcast. Users receiving se can reserve beacon for subsequent transmissions using request to send clear to send (RTS/CTS) mechanism.

6 Sensors 2017, 17, The sequence transmissions, including list channels power levels that are used during packet exchange with a beacon, is predefined fixed beforeh, all users follow same sequence. The total number exchanged to collect RSSI depends on three factors: (a) number frequency channels tested (denoted as K), (b) number different power levels considered (L), (c) retransmissions for a fixed combination a channel a power level (N). Thus, total number exchanged is: 2 N L K. After receiving a RTS or CTS packet, or beacons users stop using radio during a certain time. This period time is long enough to allow user that has reserved beacon to exchange all at first channel. A user a beacon have a fixed period to complete packet exchanging process at every different channel. When this period ends, y have to change from current channel to next. The number channels, power levels retransmissions are fixed, being configured during system initialization. These provide a trade-f between localization precision system needed to collect RSSI samples, since in general with more information a higher precision can be achieved Protocol Proposal The considered scenario includes a specific number users infrastructure beacons that broadcast periodically showing ir availability. The first user that reserves beacon using a RTS/CTS packet exchange gains channel access during a period time that is long enough to send receive N at every power level L, at first channel group K channels selected. This process can be seen in Figure 3, where a possible example is presented. In this case, beacon Bi announces its availability transmitting broadcast at channel Ch1. This is represented in Figure 3 with symbol (CSMA <Ch1>) packet <Alive,Ch1>, which indicates transmission packet Alive using CSMA protocol. After reception this broadcast packet, User2 reserves beacon Bi using RTS/CTS mechanism. This process involves transmission two RTS CTS using CSMA protocol. All rest users beacons that have received RTS or CTS start a timer with an overflow time equal to Channel Reserved time. During this period, beacons stop transmissions broadcast because channel is occupied. Next, User2 starts exchanging process with beacon Bi using channel Ch2. The sequence starts with a power level P 1 (<P 1,Ch2> in Figure 3) finishes with a power level P L. Once beacon Bi User2 have exchanged required for all power levels (from P 1 to P L ), y pass to channel Ch3 start a similar sequence. At this instant, Channel Reserved timers fire, since Ch2 has been released, or beacons (for example Bi + 1 in Figure 3) can resume announcement ir availability using Ch1. During packet exchange process with beacon Bi at different channels Ch2, Ch3,..., Chk, User2 can access channel without having to implement CSMA protocol, since none rest users or beacons located in same radio range can transmit during this period. Once Users2 finishes transmission at first channel Ch2, Channel Reserved time is over, rest users beacons restart ir radio transceivers. If any or user finds a free beacon that is prepared to exchange (for example beacon Bi + 1 in Figure 3), after RTS/CTS hshake y can start transmission in channel Ch2, since first user (User2) has already released it. Therefore, protocol takes advantage frequency multiplexing capability system when several channels have to be tested, allowing that different users exchange testing with corresponding beacons in parallel. But, to keep complexity protocol low, sequence in which channels are tested is always same.

7 Sensors 2017, 17, Sensors 2017, 17, Sensors 2017, 17, Figure Figure 3. Sequence 3. user selection channel access access in in proposed proposed protocol. protocol. In figure, In Ch1 figure, channel 1, Ch2 denotes channel 2 so on. Pi Ch1 indicates a packet Figure identifies 3. Sequence frequency user selection channel 1, Ch2 channel denotes access channel in proposed 2 protocol. so on. PIn i indicates figure, Ch1 a packet transmission with this power level. RTS CTS are transmitted to establish transmission identifies with this frequency powerchannel level. RTS 1, Ch2 CTS denotes are channel 2 transmitted so on. to Pi establish indicates a packet hshake hshake Alive is broadcast packet sent by a beacon to announce its availability. When a transmission Alive with broadcast this power packet level. sentrts by a beacon CTS are to announce its transmitted availability. to establish When a packet packet transmission is marked with CSMA term means that mote has to implement this transmission hshake is marked Alive with is broadcast CSMA term packet means sent by that a beacon mote to has announce to implement its availability. this medium When a access. medium access. Channel Reserved time is period time in which an user a beacon exchange Channel packet all Reserved transmission time at a certain is marked period with frequency time CSMA channel, in which term after anmeans expiration user that a beacon mote this time exchange has to implement y has all to go to this at a certain medium next frequency access. Channel channel. channel, Reserved after time expiration is period this time in y which has an to user go to a beacon next channel. exchange all at a certain frequency channel, after expiration this time y has to go to next channel. The channel The channel access access sequences sequences for for several users exchanging with with different different beacons beacons can can be seen in Figure 4. It should be noticed that user beacon have a fixed period to complete be seen inthe Figure channel 4. Itaccess should sequences be noticed for several that a users exchanging a beacon have awith fixedifferent periodbeacons to complete can packet exchanging process at every different channel y must change from current packet be exchanging seen Figure process 4. It should at every be noticed different that channel a user a beacon y must have change a fixed period from to complete current channel channel to next whenever this period ends. to packet next whenever exchanging this process period at ends. every different channel y must change from current channel to next whenever this period ends. RTS/CTS RTS/CTS RTS/CTS RTS/CTS RTS/CTS RTS/CTS Figure 4. Channel access sequences (Ch1, Ch2, Ch3 Ch4) for three users (IDk, IDl IDm) exchanging with three beacons (B1, B2 B3) in parallel. Figure shows how Figure Figure Channel access sequences (Ch1, Ch2, Ch2, Ch3 Ch3 Ch4) for Ch4) three forusers three(idk, users IDl (IDk, IDm) IDl exchanges between different pairs users beacons can be multiplexed in different frequency IDm) exchanging with with three three beacons beacons (B1, B2 (B1, B2 B3) in parallel. B3) in Figure parallel. shows Figure how shows how channels. exchanges exchanges between between different different pairs pairs users users beacons can beacons multiplexed can be in multiplexed different frequency in different channels. frequency channels.

8 Sensors 2017, 17, In Figure 4, user IDk exchanges with beacon B1 using channel Ch2 for all power levels considered in location process, {P j }. Once user IDk has finished sequence at Ch2, this channel is released it could be used by or users. In this case, beacon B2 initiates announcement its availability transmitting CSMA using Ch1, while beacon B1continues exchanging with user IDk, now at channel Ch3. After RTS/CTS hshake, IDl begins exchanging with beacon B2 starting at channel Ch2 following same sequence channels that B1 IDk. As it can be seen in Figure 4, our protocol allows that different pairs users beacons perform simultaneous transmissions at different channels. With avoidance contention period during transmission every testing packet, total RSSI collection decreases significantly. As it was stated in previous section, for applications programmed with TinyOS worst case T Backf is 10 ms its average value is approximately 5 ms. Reducing this protocol can speed up RSSI collection process location estimation. Indeed, this fact may be very important in tracking applications, where response time system may impose a strict refresh time Energy Consumption Analysis Energy consumption is a central problem in design WSNs. These networks are frequently deployed outdoors, in places where accessing mote to replace batteries is difficult. In se scenarios, network operation can be extended reducing mote energy consumption, which usually leads to implementation some duty cycle in radio operation. The main drawback this solution is that normally causes an increase transmission, since both motes (emitter receiver) have to wait coordinate operation periods ir wireless transceivers to perform transmission. However, localization systems present some differences in this respect. Thus, although underlying infrastructure is also based on WSN motes, y are mainly deployed indoors network motes can be easily accessed to replace ir batteries. Moreover, system beacons, which are part network infrastructure, are placed in fixed positions y could be connected to mains supply. On or h, since principal objective proposed MAC protocol is reduction RSSI collection, inclusion a radio duty cycle is not advisable. Consequently, proposed MAC protocol keeps permanently wireless transceiver in reception state. Only when a packet is transmitted, protocol changes transceiver state from reception to transmission. With this configuration radio chip, energy consumption will be higher compared to any or MAC layer protocol that establishes a duty cycle in radio operation, but keeping constantly radio in listening state significantly decreases transmission. In addition, energy consumption increase is not so critical in indoor localization systems, since mobile users can replace ir motes batteries with ease static beacons can be connected to mains supply. The energy consumption user mote can be determined considering periods time in which mote is in reception or in transmission. It is assumed that by default mote is in reception mode, unless mote is transmitting a packet. In this last case, power consumption changes accordingly to power level configured in packet transmission. Hence, energy consumption user mote for garing all RSSI can be expressed as: (( E = B T d L ) T txl ) P rx + T txl P txl L where previous terms are: (a) T d is total in RSSI collection process, (b) T txl is total time transmitting with power level L, (c) P rx is mote power consumption at reception state, (d) P txl is mote power consumption at transmission state with power level L, (e) B is number beacons that are tested in RSSI collection process. (5)

9 Sensors Sensors 2017, 2017, 17, 17, Results 4. Results 4.1. Simulation Results 4.1. Simulation Results The first evaluation considers estimation total needed to collect all RSSI samples. The first The evaluation simulated considers scenario, implemented estimationin Cooja total simulator, needed comprises to collect one allbeacon RSSI samples. one user The that simulated exchange scenario, implemented to acquire all in Cooja RSSI. simulator, Both comprises elements one were beacon implemented one user using that Telosb exchange motes to stware acquire all was programmed RSSI. with BothTinyOS. elements In were experiments, implementeddifferent using Telosb numbers motes channels stware power was programmed levels in withrssi TinyOS. sampling In process experiments, have different been considered. numbers In channels contrast, power number levels retransmissions in RSSI sampling remained process fixed have (N been = 10). considered. It should be Innoticed contrast, that number exchange retransmissions process remained always entails fixed (N = transmission 10). It should be two noticed that in opposite directions, exchangei.e., process a first always packet entails from user transmission mote to two beacon a insecond opposite packet directions, from i.e., beacon a firstto packet user frommote. user Thus, mote total number beacon a second that packet are transmitted from beacon in both to directions user is: mote. Thus, total number that are transmitted in both directions is: =2 (6) where factor 2 comes from Total Number two o f Packets transmitted = 2 N K L B in both directions, N is number (6) retransmissions, K is number channels, L is number power levels B is number where beacons. factor For example, 2 comes when from N = 10, two L = 5, K = 5 transmitted B = 1, in both total directions, number N is number is 500. In retransmissions, application K programmed is number with TinyOS channels, L send is time, number received power time, levels packet B is time number average beacons. backf For example, time were when experimentally N = 10, L = 5, measured K = 5 B = 1, ir total were number 2.2, 1.8, 0.9 is ms In respectively. application The programmed addition se with TinyOS with send rest time, terms received in Equation time, (1) gives packet time total in average transmission backf time were reception experimentally a packet measured that is approximately ir 10 were ms. This 2.2, 1.8, 0.9 multiplied 5.1 ms by respectively. number The addition provides se an estimation with rest terms total in Equation required (1) gives to collect total all RSSI in samples. transmission For previous reception example, a packet total that is approximately is in order 10 ms. five This seconds. multiplied by number Results in Figure provides 5 compare an estimation two cases: total (a) when required motes to run collect all proposed RSSI new samples. MAC For protocol, previous denoted example, as without total CSMA, is in (b) order when five motes seconds. run stard MAC protocol, described Results as in Figure with CSMA. 5 compare As two it can cases: be seen (a) when in Figure motes 5, run proposed proposed protocol new MAC without protocol, CSMA denoted shortens as significantly without CSMA, total (b) collection when motes compared run stard to case MAC with protocol, CSMA described divides as with this CSMA. time by As two. it can This be seen is due in Figure to 5, TinyOS proposed implementation protocol without CSMA CSMA shortens that includes significantly an average total backf collection time that is compared approximately to equal case with to CSMA sum all divides rest this time s by involved two. This in is a due packet to transmission. TinyOS implementation Thus, results in this CSMA Figure that includes 5 demonstrate an average backf total time reduction that is approximately that can be equal achieved to with sum all proposed rest MAC s protocol. involved in a packet transmission. Thus, results in this Figure 5 demonstrate total reduction that can be achieved with proposed MAC protocol L=10, without CSMA L=5, without CSMA L=10, with CSMA L=5, with CSMA Delay (s) Number Channels (K) Figure Delay to to collect RSSI RSSI with different number channels (K) (K) power levels (L). (L). The The number retransmissions (N) (N) at every at every combination combination channel channel power power level islevel fixedis fixed equal toequal 10. to 10.

10 Sensors 2017, 17, Sensors 2017, 17, The second assessment conducted determines aggregate network throughput achieved during The second assessment conducted determines aggregate network throughput achieved packet exchange process. Results are obtained using same previous setup simulating during packet exchange process. Results are obtained using same previous setup collection process using only one channel (K = 1). Figure 6 depicts that elimination simulating collection process using only one channel (K = 1). Figure 6 depicts that elimination CSMA increases network throughput because less time is wasted during channel access. Thus, CSMA increases network throughput because less time is wasted during channel comparing both cases it can be noticed that network throughput with new protocol duplicates access. Thus, comparing both cases it can be noticed that network throughput with new protocol throughput duplicates obtained in throughput CSMA obtained case. However, in CSMA since case. first However, stage since channel first stage reservation (based channel on reservation RTS/CTS(based hshake) on still RTS/CTS uses CSMA, hshake) impact still uses this CSMA, part is reflected impact inthis part overall is result. reflected Evenin so, overall influence result. this Even stage so, in influence total result this decreases stage in when total more result decreases when more power more levels are added more to power RSSI levels dataare garing. added to RSSI data garing. 55 Aggregate network throughput (%) N=10, without CSMA N=5, without CSMA N=10, with CSMA N=5, with CSMA Number Power Levels (L) Figure Figure 6. Aggregate 6. Aggregate network network throughput throughput with different with different power levels power (L) levels number (L) retransmissions number (N) retransmissions (10 or 5). The number (N) (10 or channels 5). The number is fixed (K channels = 1) foris all fixed combinations (K = 1) for all Lcombinations N. The aggregate L network N. The throughput aggregate network is defined throughput as percentage is defined as time percentage in which time channel in which is occupied channel with is a successful occupied packet with a successful transmission. packet transmission. Finally, last assessment presented deals with energy consumption proposed MAC protocol. Finally, The last considered assessment scenario presented is deals same with that in energy previous consumption cases energy proposed analysis MAC protocol. takes into Theaccount considered current scenario consumption is same that CC2420 in wireless previous transceiver cases [23] at energy each different analysis takes operating into account state. These current consumption are shown in Table 1. CC2420 wireless transceiver at each different operating state. These are shown in Table 1. Table 1. Current consumption CC2420 wireless transceiver at reception transmission Table modes 1. Current with different consumption power levels. CC2420 Values are wireless taken transceiver from CC2420 at reception datasheet [23]. transmission modes with different power levels. Values are taken from CC2420 datasheet [23]. Mode Current Consumption Reception Mode Current18.8 Consumption ma Transmission (L = 25 dbm) 8.5 ma Reception 18.8 ma Transmission (L 15 dbm) 9.9 ma Transmission (L = 25 dbm) 8.5 ma Transmission (L (L = 10 15dBm) ma ma Transmission (L = 7 10 dbm) ma Transmission (L = 5 7 dbm) ma Transmission (L (L = 3 5 dbm) ma ma Transmission Transmission (L (L = 1 3 dbm) dbm) ma ma Transmission (L = 1 dbm) 16.5 ma Transmission (L 0 dbm) 17.4 ma Transmission (L = 0 dbm) 17.4 ma The energy consumption analysis focuses on wireless transceiver because it is by far component The energy with consumption highest consumption analysis focuses in a Telosb on mote. wireless Assuming transceiver that because wireless it is transceiver by far component has a power with supply highest consumption 1.8V applying a Telosb Equation mote. (5), Assuming total that energy wireless consumption transceiver for hastransmitting a power supply receiving 1.8V all applying RSSI Equation can (5), be calculated. total energy Figure consumption 7 presents for transmitting obtained

11 Sensors 2017, 17, Sensors 2017, 17, receiving all RSSI can be calculated. Figure 7 presents obtained results, as it can be results, seen as it new can MAC be seen protocol new without MAC protocol CSMA outperforms without CSMA outperforms CSMA case, due CSMA to reduction case, due to reduction listening intervals listening caused intervals by elimination caused by elimination contention periods. contention It is also periods. significant It is also increase significant energy increase consumption energy with consumption number with retransmissions number retransmissions because total because number total number total in total RSSI collection in process RSSI collection rise. process rise. Energy Consumption (mj) N=10, without CSMA N=5, without CSMA N=10, with CSMA N=5, with CSMA Number Power Levels (L) Figure 7. Energy consumption for different number power levels (L). The number frequency channels (K (K = = 1) 1) beacons beacons (B (B = 1) = remained 1) remained fixed. fixed. There There were were two numbers two numbers retransmissions retransmissions (N = 5) (N or (N = 5) = 10) or (N = 810) power levels: 8 power ( 25 levels: dbm, ( dbm, dbm, dbm, dbm, 7 10 dbm, dbm, 5 dbm, 7 3 dbm, dbm, 5 1 dbm, dbm, 3 0 dbm). dbm, Power 1 dbm, levels 0 dbm). are added Power in order, levels i.e., are added case in with order, only i.e., one power case level with uses only 25 one dbm, power level case with uses two 25 power dbm, levels case includes with two transmissions power levels with includes 25 transmissions 15 dbm with so 25 on. 15 dbm so on Experimental Results The proposed protocol has been evaluated experimentally in our facilities. The experimental setup includes an area consisting three fices, a large laboratory part main corridor. The considered scenario covers a rectangular box-shaped area dimensions 9 m 16 m, which is divided by walls contains furniture, as it is shown in Figure 8. Figure 8. Covered area in experimental setup deployed in our facilities. The scenario covers three fices, a laboratory part main corridor. The The setup setup also also comprises comprises 6 6 beacons beacons that that were were deployed deployed at different at different fices, fices, in in laboratory laboratory corridor. corridor. The specific The specific location location every every beacon beacon is represented is represented by a green by a spot green in spot Figure in 8, Figure besides 8, besides red line shows path followed by a robot that was used to collect RSSI at blue spots. Figure 9 includes two pictures showing laboratory main corridor.

12 Sensors 2017, 17, red line shows path followed by a robot that was used to collect RSSI at blue Sensors spots. 2017, Figure 17, includes two pictures showing laboratory main corridor Figure 9. Pictures showing laboratory main corridor included in testing area. Figure 9. Pictures showing laboratory main corridor included in testing area. The beacons used in setup were constituted using Telosb motes. The robot was connected to an additional The beacons Telosb used mote in placed setup on were a structure constituted assembled using Telosb on motes. robot The that robot was was a meter connected high to to avoid an additional ground Telosb effect mote in placed signal on a transmission. structure assembled The collection on robot process that has was involved a meter high sampling to avoid RSSI ground effect at in every signal testing transmission. point (blue points The collection in Figure process 8) exchanging has involved with sampling 6 beacons RSSI at both sides at every (two testing point were (blue sent points in opposite in Figure directions 8) exchanging to collect RSSI with 6 beacons at at beacon both sides (two at robot). were In addition, sent in opposite different directions frequency to collect channels RSSI power at levels beacon were used at during robot). garing In addition, different RSSI frequency at every channels testing point. power levels were used during garing RSSI Once at every database testing point. was completed, a second round started to collect new RSSI for localizing Once one database user in was setup completed, area. During a second this round step started a new sequence to collect new RSSI RSSI were for localizing taken at one testing user in points. setup With area. se During, this step position a new sequence user was RSSI estimated were using taken at localization testing points. algorithm With presented se, in [24]. The position main feature user this was algorithm estimated is its using capacity to localization take advantage algorithm all presented information [24]. that The can main be drawn feature when this multiple algorithm channels is its capacity power tolevels take advantage are used during all RSSI information sampling that process. can be drawn This additional when multiple information channels can improve power levels localization are used during precision at RSSI expense sampling process. increasing This additional effort needed information to collect can improve RSSI. localization Additionally, precision localization at expense time is also increasing increased effort because needed user to collect has to transmit RSSI. receive Additionally, more localization to gar all time isrssi also increased at different because channels user has topower transmit levels. In receive this context, more new toproposed gar allmac RSSI protocol can at effectively different channels reduce localization power levels. time In this eliminating context, contention new proposed periods MACwhen protocol a large can amount effectively RSSI reduce localization taken at different time eliminating channels power contention levels periods have to when be collected. a large amount RSSI takenthe at different experimental channels evaluation power began levels considering have to be collected. impact that number channels (K) has on localization experimental precision. evaluation To this began end, considering mean absolute impact localization that number error was channels calculated (K) placing has on localization user at every precision. testing point. To thisthe end, user mean gared absolute RSSI localization error exchanging was calculated placing with 6 beacons user at every (B = 6) testing repeating point. The user transmission gared 5 RSSI times (N = 5). exchanging The position estimation with is 6based beacons (B = algorithm 6) repeating presented transmission in [8,24]. In 5 times this experiment (N = 5). Thedifferent position estimation power levels is based (L) on numbers algorithm presented channels (K) in were [8,24]. considered In this experiment to study ir different effect power in levels localization (L) precision. numbersfigure channels 10 presents (K) were error considered obtained to study testing ir only effect one power in localization level different precision. numbers Figure 10 frequency presentschannels. error obtained As it can be testing noticed onlyin one Figure power 10, level increase different numbers number frequency channels channels. improves As itlocation can be noticed precision in regardless Figure 10, increase power level. number An additional channels conclusion improves is that in location our scenario precision due to regardless diversity power obstacles level. that An prevent additional having conclusion direct line is that sight in our between scenario due user to diversity beacons, obstacles higher power that levels provide better location accuracy because with se levels larger radio ranges can be covered.

13 Sensors 2017, 17, prevent having direct line sight between user beacons, higher power levels provide better location accuracy because with se levels larger radio ranges can be covered. Sensors 2017, 17, Sensors 2017, 17, dbm -7 dbm -5 dbm Mean Absolute Mean Localization Absolute Localization Error (m) Error (m) Number Channels (K) Figure 10. Figure Mean 10. Mean Absolute Absolute 1 Localization Error Error for for different different numbers numbers frequency frequency channels (K). channels The (K). number retransmissions 1 (N = 5), power 2 levels 3 (K = 1) 4 beacons 5(B = 6) remained 6 fixed for each The number retransmissions (N = 5), power Number levels Channels (K = (K) 1) beacons (B = 6) remained fixed different line in graph. Three lines are presented according to three different power levels: for each different line in graph. Three lines are presented according to three different power levels: ( 10 Figure dbm, Mean dbm, Absolute 5 dbm). Localization Channels used Error are: for [11 different (2405 MHz), numbers 13 (2415 frequency MHz), 16 channels (2430 MHz), (K). The 19 ( 10 dbm, ( MHz), dbm, 22 ( dbm). MHz), 26 Channels (2480 MHz)]. usedchannels are: [11 are (2405 added MHz), in order, 13 i.e., (2415 case MHz), with 16 only (2430 one MHz), number retransmissions (N = 5), power levels (K = 1) beacons (B = 6) remained fixed for each 19 (2445channel different MHz), uses 22 line (2460 channel in MHz), graph. 11, 26 Three case (2480 lines with MHz)]. are two presented channels Channels according includes are added channels to three in order, 11 different i.e., 13, power case so levels: on. with only one channel Localization ( 10 dbm, uses 7 channel was dbm, carried 5 11, dbm). out after Channels case transmitting with used two are: [11 channels receiving (2405 MHz), includes all 13 (2415 channels MHz), exchanged 1611(2430 between MHz), 13, one 19 so on. Localization user (2445 MHz), was six carried beacons. 22 (2460 out MHz), after 26 transmitting (2480 MHz)]. Channels receiving are added all in order, i.e., exchanged case with only between one one user channel six beacons. uses channel 11, case with two channels includes channels 11 13, so on. The Localization second was experiment carried out was after focused transmitting on effect receiving adding all more power exchanged levels between (L). In this one case, at every user testing six point beacons. user exchanged using 6 frequency channels (K = 6) a sequence different power levels (L = [1 to 6]) with 5 retransmissions (N = 5). Results presented in Figure 11 show The that second addition experiment more was focused RSSI on collected effect adding at different more power levels (L). improves In this case, localization at every testing precision. point user exchanged using 6 frequency channels (K = 6) a sequence different power levels (L = [1 to 6]) with 5 retransmissions (N = 5). Results presented in Figure 11 show that addition 1.8 more RSSI collected at different power levels improves localization precision. 1.7 The second experiment was focused on effect adding more power levels (L). In this case, at every testing point user exchanged using 6 frequency channels (K = 6) a sequence different power levels (L = [1 to 6]) with 5 retransmissions (N = 5). Results presented in Figure 11 show that addition more RSSI collected at different power levels improves localization precision. Mean Absolute Mean Localization Absolute Localization Error (m) Error (m) -10 dbm -7 dbm -5 dbm Number Power Levels (L) Figure 11. Mean Absolute 0.9 Localization Error for different number power levels (L). The number retransmissions (N = 5), frequency 1 channels 2 (K 3= 6) beacons 4 (B 5= 6) remained 6 fixed. Channels Number Power Levels (L) were: [11 (2405 MHz), 13 (2415 MHz), 16 (2430 MHz), 19 (2445 MHz), 22 (2460 MHz), 26 (2480 MHz)]. There were 6 power levels considered: (0 dbm, 1 dbm, 3 dbm, 5 dbm, 7 dbm, 10 dbm). Power Figure 11. Figure Mean 11. Mean Absolute Absolute Localization Error for different number number power power levels (L). levels The (L). number The number levels retransmissions are added (N in order, = 5), frequency i.e., case channels with only (K = one 6) power beacons level uses (B = 0dBm 6) remained level, fixed. case Channels with two retransmissions (N = 5), frequency channels (K = 6) beacons (B = 6) remained fixed. Channels power were: [11 levels (2405 includes MHz), transmissions 13 (2415 MHz), with 16 ( MHz), 1 dbm, 19 (2445 so MHz), on. Localization 22 (2460 MHz), was carried 26 (2480 out MHz)]. after were: [11 transmitting (2405 MHz), There were 6 power receiving 13 (2415 levels all MHz), considered: 16 (2430 (0 exchanged MHz), dbm, 1 dbm, between 19 ( dbm, one MHz), 5 user dbm, (2460 dbm, beacons. MHz), 26 (2480 MHz)]. 10 dbm). Power There were levels 6 power are added levels in order, considered: i.e., case (0 dbm, with only 1 one dbm, power 3 dbm, level uses 5 0dBm dbm, level, 7 dbm, case 10 with dbm). two Power levels are power added levels inincludes order, i.e., transmissions case with 0 only 1 one dbm, power so level on. Localization uses 0dBmwas level, carried out case after with two power levels transmitting includes transmissions receiving all with 0 exchanged 1 dbm, between one so on. user Localization 6 beacons. was carried out after transmitting receiving all exchanged between one user 6 beacons.

14 Sensors 2017, 17, Sensors 2017, 17, However, improvement in location accuracy is isachieved at a expense increase in in number exchanged, which causes longer s in location time. It should be noticed that results in Figures are obtained regardless MAC protocol implemented in motes. This is because MAC protocol can change total collection, which is not represented in in se se graphs, but but acquired acquired RSSI RSSI are are same independently same independently protocol. As protocol. a result, As a localization result, localization algorithm will algorithm estimate will estimate same position same position both protocols both willprotocols provide will same provide localization same error. localization These two error. experiments These two experiments have been included have been to included highlightto highlight benefits including benefits more including channels more channels power levels inpower localization levels in accuracy. localization However, accuracy. addition However, more channels addition more powerchannels levels increases power collection levels increases. In this sense, collection application. In this sense, proposed MAC application protocol diminishes proposed thismac effect protocol allows diminishes a higher localization this effect precision allows when a higher a maximum localization in precision RSSIwhen collection a maximum process is established. in RSSI collection process is established. The last experiment relates location precision to RSSI collection. The setup is same previous case case was was conducted conducted in our in real our deployment. real deployment. In this experiment, In this experiment, number number frequency frequency channels (K channels = 6), retransmissions (K = 6), retransmissions (N = 5) number (N = 5) beacons number (B = 6) beacons remained(b fixed. = 6) In remained contrast, fixed. number In contrast, power number levels varied power fromlevels 1 to 6 varied (L = [1from to 6]). 1 As to 6 it(l can = [1 beto seen 6]). inas Figure it can 12, be seen obtained Figure results 12, establish obtained a trade-f results establish betweena trade-f in between RSSI collection in process RSSI tocollection perform process localization to perform user, localization at one specific point user, in at one considered specific point scenario, in considered location scenario, precision that location can be achieved. precision In that this can way, be achieved. new proposed In this way, MAC protocol new proposed allows MAC reduction protocol allows RSSI collection reduction eliminating RSSI collection contention periods eliminating enabling contention concurrent periods transmission enabling several users concurrent at different transmission frequency channels. several users As itat isdifferent proved in frequency Figure 12, channels. new As MAC it is protocol proved outperforms in Figure 12, MAC new MAC protocol protocol proposed outperforms in IEEE MAC protocol stard proposed divides in IEEE by two time stard needed to collect divides all by two RSSI time needed for a specific to collect combination all RSSI K, N, B for a L, specific whichcombination establishes localization K, N, B precision. L, which establishes Thus, it can belocalization concluded that precision. proposed Thus, it algorithm can be concluded reduces by that two proposed localization algorithm for reduces a certain by two localization localization precision. for a certain localization precision. Mean Absolute Localization Error (m) N=5, K=6, L=[1 to 6], without CSMA N=5, K=6, L=[1 to 6] with CSMA RSSI Collection Delay (s) Figure 12. Mean Absolute Localization Error against RSSI collection. The number retransmissions (N (N = = 5), 5), frequency channels (K = (K 6) = 6) beacons beacons (B = (B 6) remained = 6) remained fixed. fixed. Channels Channels were: [11 were: (2405 [11 MHz), (2405 MHz), 13 ( MHz), (2415 MHz), 16 ( MHz), (2430 MHz), 19 ( MHz), (2445 MHz), 22 ( MHz), (2460 MHz), 26 ( MHz)]. (2480 MHz)]. There were There 6 power were 6 levels power considered: levels considered: (0 dbm, 1 (0 dbm, 1 3 dbm, 3 5 dbm, 5 7 dbm, dbm, 7 10 dbm, dbm). 10 dbm). Power Power levels were levels added were in added order. in Delays order. were Delays measured were measured after transmitting after transmitting receiving receiving all involved all involved with onlywith oneonly userone user six beacons. six beacons Conclusions In this article, a MAC protocol adapted specifically to RSSI collection process for indoor location systems based based on on IEEE IEEE networks, networks, with multiple with multiple power levels power levels frequency frequency channels, channels, is presented. Its main advantage over or state---art approaches is that it makes use different frequency channels to avoid medium access contention periods, without having to

15 Sensors 2017, 17, is presented. Its main advantage over or state---art approaches is that it makes use different frequency channels to avoid medium access contention periods, without having to use complex scheduling algorithms or tight clock synchronization protocols. The proposed protocol establishes an initial RTS/CTS mechanism to reserve a beacon access to medium. Once a user gains medium access, it has a channel-reserved period to exchange all required at first channel without any interference from rest users or beacons. This fact allows elimination contention periods during subsequent packet exchange process. Or users can overlap ir packet exchange processes with different beacons using channels that have been tested released by previous users. The conducted experiments have included both simulated results localization one user in a real deployment. Simulated results show benefits eliminating contention periods in terms total RSSI collection network throughput. In addition, new protocol proposed is rar simple it can be easily implemented in WSNs motes with limited resources. Experimental results obtained in a real deployment with Telosb motes have proved suitability new MAC protocol for reducing RSSI collection. In addition, real deployment results provide a trade-f between location precision RSSI collection when multiple channels power levels are added. Acknowledgments: This work was supported by Spanish Government under project BIA C3-1-R by Generalitat Valeciana under project PROMETE0/2016/066. Author Contributions: Juan J. Pérez-Solano, José M. Claver Santiago Ezpeleta conceived new MAC protocol, designed experiments, performed data collecting process, analyzed data, wrote paper. Conflicts Interest: The authors declare no conflict interest. References 1. Song, Z.; Jiang, G.; Huang, C. A Survey on Indoor Positioning Technologies. In Theoretical Mamatical Foundations Computer Science; Springer: Berlin, Germany, 2011; pp Liu, H.; Darabi, H.; Banerjee, P.; Liu, J. Survey Wireless Indoor Positioning Techniques Systems. IEEE Trans. Syst. Man Cybern. Part C 2007, 37, [CrossRef] 3. Sun, G.; Chen, J.; Guo, W.; Liu, K. Signal processing techniques in network-aided positioning: A survey state---art positioning designs. IEEE Signal Process. Mag. 2005, 22, [CrossRef] 4. Deak, G.; Curran, K.; Condell, J. A survey active passive indoor localisation systems. Comput. Commun. 2012, 35, [CrossRef] 5. Bahl, P.; Padmanabhan, V. RADAR: An In-Building RF-Based User Location Tracking System. INFOCOM In Proceedings IEEE Nineteenth Annual Joint Conference IEEE Computer Communications Societies, Tel Aviv, Israel, March 2000; pp Torres-Sospedra, J.; Montoliu, R.; Trilles, S.; Belmonte, O.; Huerta, J. Comprehensive analysis distance similarity measures for Wi-Fi fingerprinting indoor positioning systems. Expert Syst. Appl. 2015, 42, [CrossRef] 7. Ezpeleta, S.; Claver, J.M.; Pérez-Solano, J.J.; Martí, J.V. RF-Based Location Using Interpolation Functions to Reduce Fingerprint Mapping. Sensors 2015, 15, [CrossRef] [PubMed] 8. Claver, J.M.; Ezpeleta, S.; Martí, J.V.; Pérez-Solano, J.J. Analysis RF-based Indoor Localization with Multiple Channels Signal Strengths. In Proceedings 8th Wireless Internet International Conference, WICON, Lisbon, Portugal, November 2014; pp IEEE Std Available online: &isnumber=27762 (accessed on 11 April 2017). 10. Gang, L.; Krishnamachari, B.; Raghavendra, C.S. Performance evaluation IEEE MAC for low-rate low-power wireless networks. In Proceedings IEEE International Conference on Performance, Computing, Communications, Phoenix, AZ, USA, April 2004; pp Petrova, M.; Riihijarvi, J.; Mahonen, P.; Labella, S. Performance study IEEE using measurements simulations. In Proceedings IEEE Wireless Communications Networking Conference, Las Vegas, NV, USA, 3 6 April 2006; pp

16 Sensors 2017, 17, Latré, B.; De Mil, P.; Moerman, I.; Van Dierdonck, N.; Dhoedt, B.; Demeester, P. Throughput analysis unslotted IEEE J. Netw. 2006, 1, [CrossRef] 13. Ergen, S.C.; Varaiya, P. TDMA scheduling algorithms for wireless sensor networks. Wirel. Netw. 2010, 16, [CrossRef] 14. Zhou, G.; Huang, C.; Yan, T.; He, T.; Stankovic, J.A.; Abdelzaher, T.F. MMSN: Multi-frequency media access control for wireless sensor networks. In Proceedings 25th IEEE International Conference on Computer Communications, Barcelona, Spain, April 2006; pp Kim, Y.; Shin, H.; Cha, H. Y-MAC: An energy-efficient multi-channel MAC protocol for dense wireless sensor networks. In Proceedings 7th International Conference on Information Processing in Sensor Networks, St. Louis, MO, USA, April 2008; pp Jovanovic, M.D.; Djordjevic, G.L. TFMAC: Multi-channel MAC protocol for wireless sensor networks. In Proceedings 8th Telecommunications in Modern Satellite, Cable Broadcasting Services Conference, Nis, Serbia, September 2007; pp Wu, Y.; Stankovic, J.A.; He, T.; Lin, S. Realistic efficient multi-channel communications in wireless sensor networks. In Proceedings 27th Conference on Computer Communications, Phoenix, AZ, USA, April Van Vinh, P.; Oh, H. Optimized Sharable-Slot Allocation Using Multiple Channels to Reduce Data-Garing Delay in Wireless Sensor Networks. Sensors 2016, 16, 505. [CrossRef] [PubMed] 19. Fonseca, J.A.; Bartolomeu, P. A MAC protocol to manage communications in localization systems based on IEEE In Proceedings th Annual Conference IEEE Industrial Electronics, Orlo, FL, USA, November 2008; pp Polastre, J.; Szewczyk, R.; Culler, D. Telos: Enabling ultra-low power wireless research. In Proceedings 4th International Symposium on Information Processing in Sensor Networks, IPSN, Los Angeles, CA, USA, April 2005; pp Osterlind, F.; Dunkels, A.; Eriksson, J.; Finne, N.; Voigt, T. Cross-Level Sensor Network Simulation with COOJA. In Proceedings 31th IEEE Conference on Local Computer Networks, Tampa, FL, USA, November 2006; pp Levis, P.; Gay, D. TinyOS Programming; Cambridge University Press: Cambridge, UK, Texas Instruments. CC GHz IEEE /ZigBee-Ready RF Transceiver Datasheet. Available online: (accessed on 25 June 2017). 24. Marti, J.; Sales, J.; Marin, R.; Jimenez-Ruiz, E. Localization mobile sensors actuators for intervention in low-visibility conditions: The ZigBee fingerprinting approach. Int. J. Distrib. Sens. Netw. 2012, 2012, [CrossRef] 2017 by authors. Licensee MDPI, Basel, Switzerl. This article is an open access article distributed under terms conditions Creative Commons Attribution (CC BY) license (