This article has been accepted for inclusion in a future issue of this journal.Content is final as presented,with the exception of pagination. XIE er al:MANAGING RFID DATA:CHALLENGES.OPPORTUNITIES AND SOLUTIONS 15 TABLE IX CRITICAL FACTORS FOR SYSTEM PERFORMANCE Scanning Range Read Throughput Energy Consumption Reader's power The effective scanning range in- If the power is too small,some The energy consumpton creases/decreases as the power is tags cannot be effectively activated increases/decreases as the power increased/decreased. or identified;If the power is too is increased/decreased. large,the interferences among tags are increased,thus the throughput is reduced. Path loss,multi- Signal attenuation is caused,thus Some tags in the normal scanning The reader has to increase the path effect,en- the effective scanning range is re. range cannot be effectively acti- power to compensate the energy ergy absorption duced. vated or identified,the throughput loss in signal propagation,the en- is reduced. ergy consumption is increased. Signal interfer- The backscattered signals of some The probability of bit errors in The reader needs to reasonably ad- ence tags cannot be effectively resolved transmitting is increased.the just the power to prevent too many due to the interference.the effec throughput is reduced. interferences,which will change tive scanning range is reduced. the energy consumption. Deployment of Dense deployment of tags will af- If the tag's plane is perpendicular Unreasonable deployment of tags tags fect the electromagnetic field of the to the incident power,the backscat- causes the reader to increase the reader's antennas,and change the ter efficiency is improved,then the power to enhance the backscattered reader's effective scanning range read throughput is increased. signal strength,thus the energy consumption is increased. Realizing that the reader's transmission power actually has among two or more protocol layers,such that the performance a significant impact on the reading performance of the RFID of network systems can be greatly improved.In the research system,a number of researchers start to investigate into this area of sensor network,the idea of cross-layer optimization area.Xu et al.investigate the impact of transmission power has been widely accepted and used.However,previously only on reading performance through extensive empirical study a few researchers leveraged this idea to solve the problems on passive tags [91][92].While exploring the relationship in RFID systems [94][95].We find that it is usually rather between the transmission power and the response quality,they difficult to solve a certain problem,e.g.,privacy protection and reveal that there exists a"lossy state"of passive tags where the authentication in RFID system,from a single layer's view. reader cannot detect tags due to insufficient power although However,given the cross-layer information,these problems tags could have responded.Based on this understanding,they can be skillfully solved by extracting and integrating the present an energy-efficient inventory algorithm.It incremen- features from various layers.For example,by correlating the tally adjusts power level to use sufficient but not excessive signal spectral features in the physical layer and the usage power for tag identification.Su et al.find that,when the pattern detected in the application layer,the specified tags can transmission power is set to a reasonable range,the"capture be effectively authenticated in RFID systems.Therefore,on effect"can be used to resolve the collision slots into singleton the one hand,if we do not sufficiently consider the influence slots [93].More specifically,the "capture effect"helps tags among different layers,the corresponding research results can- near the reader to transmit their data although collision had not achieve the expected performance in realistic applications; occurred in the time slot they used,since their signal is on the other hand,the resource scarcity of the RFID system stronger than the farthest tags due to channel attenuation. proposes the pressing needs to devise lightweight protocols Therefore,they propose a progressing scanning algorithm to and algorithms,the cross-layer optimization provides opportu- improve the reading throughput. nities to solve the traditional problems from a different aspect, Generally speaking,empirical and experimental studies in by tactfully leveraging the features from different layers. this area are relatively few.In order to guarantee the reading performance of RFID systems in real applications,the per- (2)Breaking through the traditional thinking to realize formance tuning of RFID systems in realistic settings should RFID security and privacy protection.Note that itit rather get continuous attention from researchers in academia and difficult to rely on traditional security mechanisms to guar- industry. antee the security and privacy of RFID systems.The great challenge is brought by the resource scarcity of the RFID tags,which is vulnerable to outside interferences and can VII.FUTURE RESEARCH DIRECTIONS only afford limited computation.Therefore,it is essential to Although the study and research of RFID technology has break through the traditional thinking to realize a set of novel already get much achievement,there are still a lot of challeng- security schemes for RFID systems.Fortunately,it is noted ing problems yet to be solved.In order to help researchers to that some special physical properties can be leveraged to get a better grasp of future research directions in the area implement the corresponding security schemes.For example, of RFID,we provide more insights into the future research the backscatter-based communication brings the "proximity challenges and opportunities as follows: effect",i.e.,only those tags in the vicinity of the RFID (1)Carrying out cross-layer optimization in RFID data reader can be interrogated in a real-time approach.Besides, management:The idea of"cross-layer optimization"originates the signal fingerprints in the physical layer can be used to from the protocol design of wireless network.The core idea construct the "physical one-way function"to implement the is to realize the optimized control and information exchange properties like the hash function.Therefore,how to leverageXIE et al.: MANAGING RFID DATA: CHALLENGES, OPPORTUNITIES AND SOLUTIONS 15 TABLE IX CRITICAL FACTORS FOR SYSTEM PERFORMANCE Scanning Range Read Throughput Energy Consumption Reader’s power The effective scanning range in￾creases/decreases as the power is increased/decreased. If the power is too small, some tags cannot be effectively activated or identified; If the power is too large, the interferences among tags are increased, thus the throughput is reduced. The energy consumption increases/decreases as the power is increased/decreased. Path loss, multi￾path effect, en￾ergy absorption Signal attenuation is caused, thus the effective scanning range is re￾duced. Some tags in the normal scanning range cannot be effectively acti￾vated or identified, the throughput is reduced. The reader has to increase the power to compensate the energy loss in signal propagation, the en￾ergy consumption is increased. Signal interfer￾ence The backscattered signals of some tags cannot be effectively resolved due to the interference, the effec￾tive scanning range is reduced. The probability of bit errors in transmitting is increased, the throughput is reduced. The reader needs to reasonably ad￾just the power to prevent too many interferences, which will change the energy consumption. Deployment of tags Dense deployment of tags will af￾fect the electromagnetic field of the reader’s antennas, and change the reader’s effective scanning range. If the tag’s plane is perpendicular to the incident power, the backscat￾ter efficiency is improved, then the read throughput is increased. Unreasonable deployment of tags causes the reader to increase the power to enhance the backscattered signal strength, thus the energy consumption is increased. Realizing that the reader’s transmission power actually has a significant impact on the reading performance of the RFID system, a number of researchers start to investigate into this area. Xu et al. investigate the impact of transmission power on reading performance through extensive empirical study on passive tags [91][92]. While exploring the relationship between the transmission power and the response quality, they reveal that there exists a “lossy state” of passive tags where the reader cannot detect tags due to insufficient power although tags could have responded. Based on this understanding, they present an energy-efficient inventory algorithm. It incremen￾tally adjusts power level to use sufficient but not excessive power for tag identification. Su et al. find that, when the transmission power is set to a reasonable range, the “capture effect” can be used to resolve the collision slots into singleton slots [93]. More specifically, the “capture effect” helps tags near the reader to transmit their data although collision had occurred in the time slot they used, since their signal is stronger than the farthest tags due to channel attenuation. Therefore, they propose a progressing scanning algorithm to improve the reading throughput. Generally speaking, empirical and experimental studies in this area are relatively few. In order to guarantee the reading performance of RFID systems in real applications, the per￾formance tuning of RFID systems in realistic settings should get continuous attention from researchers in academia and industry. VII. FUTURE RESEARCH DIRECTIONS Although the study and research of RFID technology has already get much achievement, there are still a lot of challeng￾ing problems yet to be solved. In order to help researchers to get a better grasp of future research directions in the area of RFID, we provide more insights into the future research challenges and opportunities as follows: (1) Carrying out cross-layer optimization in RFID data management: The idea of “cross-layer optimization” originates from the protocol design of wireless network. The core idea is to realize the optimized control and information exchange among two or more protocol layers, such that the performance of network systems can be greatly improved. In the research area of sensor network, the idea of cross-layer optimization has been widely accepted and used. However, previously only a few researchers leveraged this idea to solve the problems in RFID systems [94] [95]. We find that it is usually rather difficult to solve a certain problem, e.g., privacy protection and authentication in RFID system, from a single layer’s view. However, given the cross-layer information, these problems can be skillfully solved by extracting and integrating the features from various layers. For example, by correlating the signal spectral features in the physical layer and the usage pattern detected in the application layer, the specified tags can be effectively authenticated in RFID systems. Therefore, on the one hand, if we do not sufficiently consider the influence among different layers, the corresponding research results can￾not achieve the expected performance in realistic applications; on the other hand, the resource scarcity of the RFID system proposes the pressing needs to devise lightweight protocols and algorithms, the cross-layer optimization provides opportu￾nities to solve the traditional problems from a different aspect, by tactfully leveraging the features from different layers. (2) Breaking through the traditional thinking to realize RFID security and privacy protection. Note that it it rather difficult to rely on traditional security mechanisms to guar￾antee the security and privacy of RFID systems. The great challenge is brought by the resource scarcity of the RFID tags, which is vulnerable to outside interferences and can only afford limited computation. Therefore, it is essential to break through the traditional thinking to realize a set of novel security schemes for RFID systems. Fortunately, it is noted that some special physical properties can be leveraged to implement the corresponding security schemes. For example, the backscatter-based communication brings the “proximity effect”, i.e., only those tags in the vicinity of the RFID reader can be interrogated in a real-time approach. Besides, the signal fingerprints in the physical layer can be used to construct the “physical one-way function” to implement the properties like the hash function. Therefore, how to leverage This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination