History

A novel paradigm to exchange data in RFID piconets

Description
A novel paradigm to exchange data in RFID piconets
Categories
Published
of 8
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
Related Documents
Share
Transcript
   A novel paradigm to exchange data in RFID piconets   I. Farris, A. Iera, and S.C. Spinella Dept. DIIES - University Mediterranea of Reggio Calabria A.R.T.S. Laboratory (http://www.arts.unirc.it), Loc. Feo di Vito, 89100, Reggio Calabria, Italy ivan.farris.524@studenti.unirc.it, antonio.iera@unirc.it, silverio.spinella@unirc.it  Abstract   —In this paper a new paradigm is proposed, according to which a group of RFID readers establishes a piconet, called RAN (RFID Area Network), similarly to Bluetooth or Zigbee devices and exchanges data by only using RFID tags as a common “virtual channel”. The proposal represents an interesting enabling factor of  pure  RFID ecosystems, wherein RFID enabled devices (such as mobile RFID readers, RFID reader cards embedded into cellular/mobile devices, etc.) only rely on the RFID technology for identification, for sensing and, now, for low-bitrate data exchange as well. Suitable algorithms to handle interference and collision problems during the data exchange within a piconet are proposed and the performance of the introduced paradigm is assessed.  Keywords—RFID virtual channel; RFID piconets; Anticollision. I.   I  NTRODUCTION  Unconventional uses of the RFID technology have recently  been investigated, which go well beyond the mere identification. In the literature and in real applications it is easy to find RFID devices used, for example: to be coupled with (or acting as) sensing devices [1], to perform indoor localization and tracking [2] [3], to behave as an out-of-band wake-up channel for sensor networks [4], etc. This multi- purpose aspect of the RFID technology makes one confident that in the future Internet of Things (IoT) it will keep on  playing a relevant role, together with sensors and actuators. To strengthen this idea, recently it has been observed an increase in the deployment of so called RFID ecosystems in different areas of everyday life [5]. In these settings, RFID technology is always used in conjunction with other technologies as for the aspects connected to the exchange of data. In paper [6] and in a patenting activity [7] we have envisaged the possibility of: (i) establishing   a  piconet called  RFID Area Network (RAN)   made up of readers, whose functions have been specially augmented to allow unicast and broadcast communications, and (ii) enabling data exchanges among the involved reader devices by only using the residual memory of passive tags  shared in a given area. In that paper we introduced the  paradigm and performed initial theoretical studies on the achievable performance level in terms of data rate in an ideal scenario wherein only two readers exchange data through a single tag. In the present paper, instead, we aim at conducting a more comprehensive performance assessment study by also addressing interference and collision problems in more realistic situations of a RAN with more than two readers. Possible uses of the proposed paradigm can be manifold. The only limitation is the reduced throughput that can be supported over the defined “virtual RFID channel” by today’s technology.  Notwithstanding, this is not a real issue if one accounts for the huge current interest of manufacturers in the RFID area and the consequent rapid advancements observed. It is easy to envisage that this paradigm can effectively be used to increase the accuracy of indoor positioning and location tracking   techniques based on RFID. One can think to indoor localization techniques based on low-cost readers (maybe simple interfaces embedded in domestic robots or devices) that may achieve an increased accuracy through data exchanges (cooperative  positioning) by only using the RFID tags around them! Also, in a scenario of distributed sensing  , low-cost miniaturized RFID (single technology) readers, distributed in the environment or embedded in objects, can gather data from sensorized RFID tags and exchange them up to reach a higher complexity RFID reader that acts as an “anchor” to the Internet (similar to traditional sensor networks [8]). Likewise, distributed search  engines for things in a given area can be implemented by means of a few fixed multi-technology RFID readers and several low-cost, mobile, single-technology (RFID only) readers moving around in the environment.  Home automation  and monitoring   applications or assisted living   and telemedicine  services are among the candidates to take full advantage of these new opportunities, as well as robot-to-robot   and machine-to-machine  low-cost communications. The persistence of the medium, based on the RFID tag memory, is a key feature of this protocol, introducing new possible application domains and innovative solutions to face several problems in the literature. For example, devices exchanging data in a given area can find themselves momentarily out of the communication range. In these cases a direct communication (whatever the used technology) would be lost. The persistence of the RFID memory and the establishment of a RAN would undoubtedly have a beneficial effect. On the other hand, small low-powered mobile readers based on the Gen2 standard, will be likely conceived so to use a sleeping mode for energy saving. In this view, the persistence of the distributed memory of the RFID in the environment can represent a novel approach to solve the well-known issue of synchronization between activity and sleeping phases typical of the devices in any sensor network. The results found testify to the great potentials of such a novel way to look at the RFID tag memory as a virtual channel in a RFID ecosystem for the future IoT. Our solution shall not be considered a direct competitor of existing wireless short range technologies, but it represents a new paradigm of communication that will be used in “almost real time” fashion or in “delay tolerant” modality by exploiting the persistence of the RFID memory, depending on the application chosen. Besides, the solution is designed to be fully compliant with the 978-1-4673-5750-0/13/$31.00 ©2013 IEEE2013 IEEE International Conference on RFID 215  EPC standard and EPCglobal platforms, allowing low-cost and low-powered RFID micro-reader, characterized to exchange data without additional hardware. Surely, a data storage into RFID tags increases delays and tag memory occupation, but the advantages in terms of overall flexibility in the RAN composition and management largely overcome the limitations. In the remainder of the paper, we first describe the main features of the novel paradigm. Then, we address the interference issues raising in a RAN and propose two methods to overcome the observed limitations. Last, a comprehensive  performance assessment campaign is reported, to point out actual potentials and limitations. II.   T HE PROPOSED PARADIGM  Fig. 1 shows the basic reference scenario for our proposal. It foresees the presence of four main entities: (1)  Master Reader (MR), in charge of creating and handling one or more  RFID  Area Networks ; (2) Client Readers (CRs), exchanging information with a MR within a RAN; (3) passive  RFID tags  of type EPCglobal Class1 Gen2, equipped with the User Memory (UM)   Bank    and used as a low-cost communication channel; (4)  Master Control (MC), an additional device associated to the MR that implements a NAT (Network Address Translation)  protocol to map IP addresses onto ID-Master addresses. Fig. 1.   Basic reference scenario CRs and RFID tags can be associated, at the same time, with different RANs, created by any MR. Suitable  Discovery Tables  are implemented in both the MR and CR to dynamically identify the RFID tags that each MR can utilize as a communication channel to transfer data towards a CR and vice-versa. Logically, to implement the whole set of procedures that our proposal foresees, the RFID UM bank must be properly structured. In order to guarantee interoperability with ISO/IEC 15962 standard, which rules the coding and decoding  procedure for the UM bank, a novel  Data Format  – described as “Communication Channel”, along the lines of ISO/IEC 15961 – and a novel organization of the UM   are proposed. In Fig. 2, the reference Protocol Stack is depicted. Full compatibility with the radio interface of the EPCglobal UHF Class1 Gen2 standard [9] is maintained and some functionalities are introduced into the higher layers: (i)    RAN establishment and maintenance sublayer  : defines the new data structures and UM format so to implement the three phases of the proposed solution:  Addressing  , Communication  and Control  ; optimizes the tags’ memory capacity usage and, at the same time, guarantees fairness in the sharing of the tags among the readers; (ii)    RAN MAC sublayer  : defines the new radio channel access schemes accounting for the interference problems raised by the presence of MR and CRs within a given RAN (  LBT-based   and  LBT-based with EMEB  policies evaluated later). Fig. 2.   Protocol Stack proposed  A.    Addressing Phase During this phase, a piconet is established. The MR acquires an  ID Master   address from the MC to which it is connected through a fixed or a traditional wireless link. As a next step, the MR performs the inventory of the tags falling within its reading range to identify which of them can be used to establish the RFID communication channel. Specifically, it has to understand if these have a UM and are not already used for any other specific application foreseen by the ISO/IEC 15961 standard. This is done by evaluating the UMI field in the Protocol Control section of the EPC Bank and checking that UM is present and is still in the uninitialized state. The DSFID (Data Storage Format Identifier) of the chosen tag is set to the value corresponding to “Communication Channel” and the remainder of UM will then be organized according to the scheme in Fig. 3. The basic hypothesis driving the specified framing is that the number of bits of the UM is a multiple of 128. In Fig. 3 it can be observed that a subset of AIs (Application Identifiers), defined Cluster Packed Object  s (CPOs), repeat themselves. The number of CPOs is equal to n (this number depends on the  User Memory Size  value in the TID Bank). A CPO can be, thus, considered like a 117 bit frame within a multi-frame structure. Fig. 3 also shows that for each CPO a Control Packet   and a  Data Packet are defined. The fields of the Control Packet   are used for this phase and the Control Phase, the  Data Packet fields, instead, are only used in the Communication Phase. Once a tag used as a virtual communication channel is identified, each CR can then be associated with one or more RANs obtaining, through the Control Packets , an address "RAL" stored into its own and the MR’s Discovery Tables and used during the Communication Phase. The Addressing phase through the exchange of Control Packets allows any CR to join a particular RAN as soon as it wins the contention and gains access to the medium.  B.   Communication Phase When the Addressing Phase is over, the Communication Phase can start to allow data traffic exchange between MR and CR. Both unicast   (i.e., an exchange of messages between the MR and a single CR) and broadcast   (i.e., an exchange of messages 2013 IEEE International Conference on RFID 216  sent by the MR to all the CRs listening to the same tag) communications are foreseen. In the unicast   case, the involved CR, for each received message, will also release the memory resources engaged on the tag (this also works as a kind of acknowledgment mechanism). As it is likely that the exchange of a message will involve more tags, then the CR may be required to release resources on several tags. In the broadcast   case, it is not possible to release the engaged resources following a read operation by the CR because it does not know which other CR has already read the data on the tag/tags (broadcast transmission without explicit acknowledgment). In the analysis of this paper a single  Priority Level   is considered and a Round Robin scheduling algorithm both for the CPO assignment and for the sharing of each RFID tag is assumed. Besides, our assessment study assumes that reliable unicast communications are performed (i.e., packets are suitably acknowledged by the CRs). This is simply performed by slightly modifying the AI Cluster Map . Fig. 3.   Structure of the UM C.   Control Phase During this phase, the Discovery Tables are updated. It can be started both by either a MR or a CR to check for the presence of a given MR/CR/Tag within a RAN over the time. Therefore, three cases can be considered: (i)  control phase verified by the MR, (ii)  control phase verified by the CR and (iii)  control phase with tag verification. III.   I  NTERFERENCE PROBLEMS  Our proposal is based on three main assumptions: (i)  the involved readers can operate on one or more shared tags (i.e., coverage areas of readers partially or totally overlap), (ii)  within the overlapping areas there are a number of tags with the available resources to implement the “virtual communication channel” and (iii)  there are no readers external to the considered RAN. Under these assumptions, the RFID readers can mutually interfere during the phases of inventory and access to the tags. In the literature the reader collision problem is investigated and two kinds of interference are defined [10]: •    Reader-to-reader interference : it happens when two readers operate at the same frequency. The reply signals received  by a reader from the tags of interest is covered by a stronger signal from an interfering reader. This is highly probable, due to the lower power (compared to the readers’ ones) at which messages are backscattered by passive RFID tags; •    Multiple-reader-to-tag interference : it happens when a tag is in the coverage areas of more readers (i.e. the condition to establish a RAN). Whenever more readers try to simultaneously query a tag, this latter is unable to reply. In some extent, the reader-to-reader interference can be addressed thanks to the dense reader modality foreseen by the EPCglobal Gen2 standard, according to which reader and tag transmit over separate (and well defined) channels. Multiple-reader-to-tag interference still exists whenever more readers transmit simultaneously and the tag is not able to decode the collided signals. The literature makes available proposals of centralized   [11] [12] [13] [14], distributed   [15] [16] [17] [18] [19] [20], and hybrid   methods [21] to face the cited interference  problems. Centralized methods are unsuited for the distributed solution we are looking for. Any distributed method proposed foresees “extra RFID technology” control and synchronization channels; this contrasts our idea of a channel for a pure RFID ecosystem. Hence the need to define a new MAC. Reader-To-Reader and Reader-To-Tag interferences are avoided by proposing an algorithm of the CSMA family, which is derived from the LBT (Listen Before Talk) [22] modality and is hereinafter referred to as  LBT-based  . It is assumed that all the readers in the RAN operate over the same unique radio channel. Before starting any transmission, each reader monitors the channel for a time interval lasting 5 ms plus a random time   value uniformly distributed in the [0ms, 5ms] interval (each time slot   duration is 0.5 ms) to detect possible transmissions ongoing. If the channel is busy, the reader continues to sense the channel until this becomes idle and then a further medium access procedure can be implemented according to the above illustrated time intervals. The ETSI standard also foresees that a reader cannot occupy the channel for longer than 4 seconds and that, once it releases the channel, it cannot try to access again before 100 ms. This last condition is disadvantageous in the case of a RAN with only two CRs; thus, it was decided to reduce it to only 10 ms. In this early stage of the research, it is  possible to assume that, within a RAN, each reader can listen any other reader, thus excluding the hidden terminal problem as evaluated in [23]. We assume that the RFID readers are equipped with omnidirectional antennas and their transmission  power is the same. Obviously, this simple  LBT-based   algorithm will introduce a large number of collisions and a high  percentage of useless accesses due to the continuous readers-to-tags polling (to verify the presence of interesting messages in the UM). To resolve the problems concerned, we introduce into the  LBT-based   algorithm a further anti-collision mechanism that we call  EMEB, Echo Message EPC Bank  . We use as a starting point the CSMA protocol (  LBT-based  ), according to 2013 IEEE International Conference on RFID 217  which the reader must always listen before transmitting over the channel to check if it is already occupied. A reader, which is monitoring the channel, can hear and decode response messages from tags that are interrogated by another reader, as in [24]. The MR writes signaling messages (the decided access sequence for the CRs) into the memory of the tag and then it  performs a read operation. In this way, through the response message, the tag can propagate the signaling information it contains to the listening readers. By doing so, the MR can define the order in which each reader has to access the transmission means, and so the tags, according to our  paradigm. This allows a reader to avoid continuously reading the UM of the tag and blindly looking for new messages. According to the  LBT-based with EMEB  algorithm, the first CR in the Echo Message sequence has no need to wait for the time interval specified by  LBT-based  , but it can access the radio channel after a time slot. Whenever a CR accesses the medium, it makes the proper read and write operations relevant to the CPO of the UM allocated to it, cancels its medium reservation in the EPC Bank, and reads again the EPC Bank to notify the next scheduled reader about its turn to access the tags. The  process continues as long as all CRs have had access to the tag and there are no more reservations in the EPC Bank. To allow a new CR to join a RAN already established, the “pure”  LBT-based   procedure is briefly activated after the last CR’s access,  before any new transmission round. In the analysis presented in this paper only static RANs are considered (i.e. no new CRs enter the RAN). To implement the protocol, all the readers shall be able to distinguish signaling messages from mere UM reading commands. The idea is to use the EPC Bank as the memory space where to put signaling messages into. This way, the tags used as a communication channel are uniquely identified within the Discovery Tables. In fact, despite the standard suggests to use EPC codes of 96 bits, the EPC Bank of several commercial tags has a dimension considerably greater, in the order of 512  bits. Signaling messages can thus be appended to the EPC Code (refer to Fig. 4). Please note that the EPC Code, whose length is defined in the Protocol Control section, remains unchanged when the same tag is used for other applications. Reading the EPC Bank will allow a reader to send the signaling message to all other listening readers. The tag reply has a reduced transmission range and, thus, it could happen that not all the readers in the RAN are able to listen to it.  Notwithstanding, the MR is able to derive from the data in the Discovery Table which readers of a RAN are able to listen to the communications of any given tag. To any CR the triplet  IDRS-IDRD-IDRAN   is associated, whose order is defined by starting from the 97 th  bit. Fig. 4.   Echo Message EPC Bank IV.   P ERFORMANCE ASSESSMENT  The EPCglobal Gen2 standard [9] regulates the interaction  between interrogator and tags through three procedures: Select  ,  Inventory , and  Access (each one comprising one or more commands). When the Inventory phase is successfully concluded, Read and Write commands can be sent. A Read operation allows to read up to 256 16-bit words (512 bytes) at a time. A Write operation, instead, allows to write a single 16-bit word at a time. One has to notice that the standard foresees a temporal window, lasting for a maximum of 20 ms, which follows any Write (or BlockWrite) operation. During this time window, the reader energizes by CW (Continuous Wave) the tag to enable the writing onto its memory. This time interval is tightly related to the tag type and, for EEPROM memories, it ranges from 4 ms to 10 ms. If a reader does not receive a tag reply in 20 ms, then the command is considered as failed. In [6] it has been evaluated the  Bit Rate [bit/s]   and    Data Rate [bit/s]  that is theoretically possible to achieve when two readers within a RAN exchange data over a single tag in ideal conditions. Three tag classes are considered: (i) F-RAM Tag  , high innovative and characterized by T Write ~  0 ms, (ii)  EEPROM Tag  , products of commercial strip, characterized by T Write = [4÷10] ms, (iii) Standard Tag  , a virtual tag characterized by the maximum parameter value T Write = [20] ms defined by the standard. We will consider only tags with a memory based on F-RAM (Ferroelectric Random Access Memory) technology (such as the MaxArias ®  products by Ramtron [25] or FerVID Family TM  products by Fujitsu [26]). Such a new type of tag allows to reduce the write time intervals, increase the rewrite capability, and equal the reading and the writing range. Whenever a reader successfully occupies the radio channel, it can access sequentially all the tags within its communication range without Tag-to-Tag collisions problems. In this case the only delay considered before accessing to UM is the one required to obtain the EPC code of a single tag ( Single Tag  Reply , like in Fig. 5). Besides the Single Tag Reply , the EPCglobal Gen2 standard also foresees the cases of: (i) Collided Reply , (ii)  No Reply  and (iii)  Invalid ACK  . We assume that after having had access to the medium each reader sequentially and exclusively accesses all the tags, used as a virtual channel, in its communication range (via the Select command, by indicating their EPC codes memorized in the Discovery Table) and that the BER on the channel is zero. Thus, we can neglect the cases of Collided Reply  (due to the Q- protocol) and  Invalid ACK. Fig. 5.   Reader – Tag communication timing diagram (Single Tag Reply) Whenever the reader sends the specific query command to access the selected tag and does not receive any response in a time interval T1+T3  (T1: time from interrogator transmission to tag response; T3: time, after T1, an interrogator waits before it issues another command), it can assume that another reader is transmitting at the same time. Therefore, the  No Reply  case can  be considered as a case in which collision event happens, as illustrated in Fig. 6. This last assumption holds when we assume, like in this paper, that there are no tags leaving the 2013 IEEE International Conference on RFID 218  RAN (static RANs). In future analyses the possibility that a tag leaves the RAN will be better accounted for. In this early paper we can assume that the unavailability of a tag is handled during the Control Phase and thus only the tags actually available in a RAN are always queried. Fig. 6.   Collision Event (No Reply) As a consequence, the collision time is computed as: Collision time = T  5 +Select+T  4 +Query+T  1 +T  3   The EPCglobal Gen2 physical layer has the dual purpose of maximizing harvestable power at the tags, and facilitating downlink and uplink communication. Depending on the coding, downlink rates (  Reader-to-Tag Data Rate ) range from 27 kbps to 128 kbps, while uplink data rates ( Tag-to-Reader  Data Rate ) are between 5 kbps and 640 kbps. In our  performance evaluation study, the main parameters of the EPCglobal Gen2 standard are set to the values listed in Table I. In the Table four configurations are considered: (i)  Maximum , identifying an upper bound configuration, (ii)  Medium 1 , identifying a configuration with a medium-high bit-rate, (iii)  Medium 2 , identifying a configuration with a medium-low bit-rate, (iv)  Minimum , identifying a lower bound configuration. A further assumption is that a MR sends messages to each CR by using one or more UM CPOs associated to a tag. Furthermore, the RFID tag response times to the Read and Write commands and the reader elaboration times are neglected. TABLE   I. P ARAMETER S ETTING   Parameters Max Med 1 Med 2 Min Tari [ µ s] 6.25 12.5 12.5 25 Data1 [ µ s] 9.375 24.375 24.375 50 Divide Ratio (DR) 64/3 64/3 64/3 8 Reader-to-Tag Data Rate [kbps] 128 54.23 54.23 27 RTcal [ µ s] 15.625 36.875 36.875 75 TRcal [ µ s] 33.3 83.3 83.3 200 BLF [KHz] 640 256 256 40 Encoding FM0 FM0 Miller (4) Miller (8) TRext 1 1 1 1 Tag-to-Reader Data Rate [kbps] 640 256 64 5 T1 [ µ s] 13.625 39 39 252 T2 [ µ s] 4.6875 39 39 250 T3 [ µ s] 18.75 39 39 250 T4 [ µ s] 31.25 73.75 73.75 225 T5 [ µ s] 1500 1500 1500 1500 The conducted simulation campaign (by a discrete-event simulator) aims at assessing the performance level of the envisaged system mainly in terms of goodput-delay tradeoff.  A.    Performance under ideal conditions For the sake of completeness, we briefly give in Fig. 7 and 8 the results of a theoretical study, in which we assumed that a MR sends messages to a single CR by using the whole UM of a single tag and the system is under conditions of perfect synchronism (ideal conditions to compute a theoretical upper  bound for the rate related metrics). It emerges that an increase in the UM corresponds to an increase in the goodput, i.e. the rate of useful information [bit/s], which can be computed as:      where (i)  BitPayload   is composed of 76 bits according to Fig. 3, (ii)  NumberCPO  is equal to the size of UM divided by 128, (iii) TimeCycle  is the summation of the time necessary to MR to put into the UM (i.e., into the different CPOs) the information for the CR ( T   MSG ) and the time that CR needs to acknowledge the message reception through the only updating of the ClusterMap ( T   ACK  ). T   MSG  e T   ACK   are modeled according to the EPCglobal Gen2 standard, comprising the procedures of Select, Inventory, Access, whose transmission time intervals relevant to the messages, sent over the wireless channel by both readers and tags, are determined using the parameters reported in Table I. Fig. 7.    Network Goodput, theoretic study  Notwithstanding, also the delay, i.e. the time between two subsequent transmissions for the same CR [ms], increases due to the longer tag reading and writing time. For low Reader-to-Tag and Tag-to-Reader data rates, like for example those chosen in the Minimum configuration, the radio channel occupation time of a reader could exceed the maximum threshold set by the ETSI standard, in case of big dimensions of UM. By starting from this analysis, it is possible to determine the UM size to meet the QoS constraints of a particular application in terms of goodput and delay. Most interesting is to evaluate, in the present paper, the  performance of a RAN composed of several CRs and more than a single RFID tag, therefore including interference and collision issues. The curves reported in the following sections are achieved by a discrete-event simulator, which models static RANs. Considering the configuration labeled 2013 IEEE International Conference on RFID 219
Search
Similar documents
View more...
Related Search
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks