CoChat
Sign Up
Cited by 125IoT and Edge/Fog ComputingCryptographic Implementations and SecurityAdvanced Authentication Protocols Security

A Practical Evaluation of a High-Security Energy-Efficient Gateway for IoT Fog Computing Applications

Open access
Suárez-Albela, Manuel, Fernández-Caramés, Tiago, Fraga-Lamas, Paula, Castedo, Luis·2017·Sensors
DOI 10.3390/s17091978PMC PMC5620735
Open this paper in CoChat — free accountOpen PDF

Abstract

Fog computing extends cloud computing to the edge of a network enabling new Internet of Things (IoT) applications and services, which may involve critical data that require privacy and security. In an IoT fog computing system, three elements can be distinguished: IoT nodes that collect data, the cloud, and interconnected IoT gateways that exchange messages with the IoT nodes and with the cloud. This article focuses on securing IoT gateways, which are assumed to be constrained in terms of computational resources, but that are able to offload some processing from the cloud and to reduce the latency in the responses to the IoT nodes. However, it is usually taken for granted that IoT gateways have direct access to the electrical grid, which is not always the case: in mission-critical applications like natural disaster relief or environmental monitoring, it is common to deploy IoT nodes and gateways in large areas where electricity comes from solar or wind energy that charge the batteries that power every device. In this article, how to secure IoT gateway communications while minimizing power consumption is analyzed. The throughput and power consumption of Rivest–Shamir–Adleman (RSA) and Elliptic Curve Cryptography (ECC) are considered, since they are really popular, but have not been thoroughly analyzed when applied to IoT scenarios. Moreover, the most widespread Transport Layer Security (TLS) cipher suites use RSA as the main public key-exchange algorithm, but the key sizes needed are not practical for most IoT devices and cannot be scaled to high security levels. In contrast, ECC represents a much lighter and scalable alternative. Thus, RSA and ECC are compared for equivalent security levels, and power consumption and data throughput are measured using a testbed of IoT gateways. The measurements obtained indicate that, in the specific fog computing scenario proposed, ECC is clearly a much better alternative than RSA, obtaining energy consumption reductions of up to 50% and a data throughput that doubles RSA in most scenarios. These conclusions are then corroborated by a frame temporal analysis of Ethernet packets. In addition, current data compression algorithms are evaluated, concluding that, when dealing with the small payloads related to IoT applications, they do not pay off in terms of real data throughput and power consumption.

AI snapshot

Written from the paper’s full text.

TL;DR

This paper evaluates TLS cipher suites for resource-constrained IoT fog gateways, comparing RSA and ECC in a real testbed. It finds ECC/ECDSA reduces energy consumption and increases throughput relative to RSA, while GZIP compression is generally not beneficial for small IoT payloads.

Key findings

  • ECC/ECDSA outperformed RSA: server-side power consumption with RSA doubled that of ECC, client-side ECC had 15–20% lower power consumption, and total energy for a 512-byte payload with 32 concurrent users was reduced about 50% using ECDSA.
  • Throughput: ECC handled more than twice the requests per second of RSA once minimal concurrency was reached; an RSA transaction took approximately 0.012 s versus 0.005 s for ECC (2.5× faster), and ECC required 2–2.5× less energy per request.
  • GZIP: for a 256-byte payload on Apache server side, compression consumed 1.9% more for RSA, 4.3% more for ECDSA, and 83.7% more for HTTP; for payloads of 128 bytes and less, compression increased transmitted data, and for 128 KB payloads no compression was faster by up to 46%.
  • Frame analysis: for a 512-byte payload, RSA transmitted about 15% more data than ECC, while ECC transmitted about 130% more data than plain HTTP; the TLS handshake took longer for RSA, with data transmission time unaffected.

Methods

  • Testbed: three Orange Pi PC single-board computers (Allwinner H3 quad-core Cortex-A7 1.6 GHz, 1 GB DDR RAM, 100 Mbit Ethernet) running ARMbian; Apache2 and Nginx with OpenSSL; Siege benchmark; external Adafruit INA219 for power measurement; Wireshark for Ethernet frame analysis.
  • Compared ECDHE-RSA-AES128-GCM-SHA256 (2048-bit RSA certificate) versus ECDHE-ECDSA-AES128-GCM-SHA256 (secp256r1/256-bit ECC certificate), plus plain HTTP baseline; payloads ranged from 32 bytes to 131,072 bytes in base-2 progression; 200 requests per client; concurrent clients from 2 to 128 in base-2 progression; 91 tests per cipher suite and web server, totaling 1092 Siege runs, with and without GZIP.
  • Analysis measured server- and client-side energy consumption, requests per second, requests/s/mWh, and Ethernet frame timing/data counts.

Limitations

  • The paper states that only server authentication was used; client authentication may be needed and would make server- and client-side energy consumption similar. It also notes that a different SBC or SoC could change total power consumption results, though cipher-suite comparison conclusions should not change, and that Wireshark frame times are approximations.
  • The paper reports an unexplained Apache performance drawback/delay in the TLS handshake with ECDSA for 512-byte payloads; the specific cause remains unknown.
  • Evident from the text: tests use wired Ethernet and SBC gateways rather than actual battery-operated or wireless IoT nodes; hardware crypto acceleration was disabled; and only two cipher suites, one ECC curve, and one RSA key size were tested, so generalization to other platforms, curves, or key sizes is not directly shown.

Model-written summary. Check it against the paper before relying on it.

Integrity signals

Registry

Crossref · 10.3390/s17091978

Cites

79 of 79 references verified0 citation errors · 0 not checkable

Statistics

Nothing to recompute.

Claims the paper makes

0 of 42 cited sources checked · 0 supported · 0 flagged

  • Findingown data

    When dealing with the small payloads related to IoT applications, current data compression algorithms do not pay off in terms of real data throughput and power consumption.

    “In addition, current data compression algorithms are evaluated, concluding that, when dealing with the small payloads related to IoT applications, they do not pay off in terms of real data throughput and power consumption.”
    Quote in the paperNot a substantial claim
  • Findingown data

    Even the less powerful boards available nowadays greatly exceed the memory values defined by each of the three classes of constrained devices.

    “As it can be seen in Table 1, even the less powerful boards available nowadays greatly exceed the memory values defined by each of the three classes.”
    Quote in the paperNot a substantial claim
  • Findingown data

    The hardware and software of the testbed has a very stable base power consumption.

    “After analyzing these results, it can be concluded that the hardware and software of the testbed has a very stable base power consumption.”
    Quote in the paperSubstantial claim
  • Findingown data

    Power consumption remains stable when the board is idling, with minimal and low amplitude deviations.

    “As it can be observed, power consumption remains stable when the board is idling, with minimal and low amplitude deviations.”
    Quote in the paperSubstantial claim
  • Findingown data

    The effect of large payloads on energy consumption does not depend on TLS or on the cipher suite selected.

    “Furthermore, by observing the HTTP energy consumption curve of Figure 10, it can be concluded that the effect of large payloads on the energy consumption does not depend on TLS or on the cipher suite selected.”
    Quote in the paperNot a substantial claim
  • May combine two claims. 63
  • Classified as probably not a substantial claim. 42

  • May combine two claims.

    Findingown data

    In the specific fog computing scenario proposed, ECC is a much better alternative than RSA, obtaining energy consumption reductions of up to 50% and a data throughput that doubles RSA in most scenarios.

    “The measurements obtained indicate that, in the specific fog computing scenario proposed, ECC is clearly a much better alternative than RSA, obtaining energy consumption reductions of up to 50% and a data throughput that doubles RSA in most scenarios.”
    Quote in the paperSubstantial claim
    • The authors state this as an assumption: IoT gateways are constrained in terms of computational resources.

  • May combine two claims.

    Findingreasoning

    When using HTTP, power consumption increases, but presents several drops that even reach idle power consumption values.

    “When using HTTP, power consumption increases, but it presents several drops that even reach idle power consumption values.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingreasoning

    At the client side, the difference in energy consumption between GZIP and no compression starts to be noticeable when the payload size exceeds 8192 bytes, although the absolute difference is smaller than in the server case.

    “In this case, the difference starts to be noticeable when the payload size exceeds 8192 bytes, although the absolute difference is smaller than in the server case.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    For a 256-byte payload with Apache, compression makes the system consume 1.9% more in the case of RSA, 4.3% more in the case of ECDSA, and 83.7% more in the case of HTTP.

    “For example, in the case of using Apache as a server, Table 3 shows, for a 256-byte payload, that compression makes the system consume 1.9% more in the case of RSA, 4.3% more in the case of ECDSA, and 83.7% more in the case of HTTP.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingreasoning

    HTTPS power consumption is the highest and presents less amplitude variation.

    “As expected, HTTPS power consumption is the highest and it presents less amplitude variation.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingown data

    The gains from GZIP are minimal and only present at the client side.

    “In any case, the gains are minimal and only present at the client side.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingown data

    For payloads of 128 bytes and less, using GZIP compression will increment power consumption and increase the size of the original data, resulting in more data being transmitted through the network.

    “For payloads of 128 bytes and less, using GZIP compression will not only increment power consumption, but it will also increase the size of the original data, resulting in more data being transmitted through the network.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingreasoning

    GZIP decreases the energy efficiency for large payloads and does not enhance it for smaller payloads.

    “To sum up: GZIP decreases the energy efficiency for large payloads and does not enhance it for smaller payloads.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    Once a minimal number of concurrent clients is reached, in this case 8, the server is constantly responding to requests, causing requests per second to remain constant even if the number of concurrent clients is increased.

    “Once a minimal number of concurrent clients is reached, in this case 8, the server is constantly responding to requests. This causes the requests per second to remain constant even if the number of concurrent clients is increased.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingown data

    For 128 kilobyte payloads, ECDSA greatly reduces the total transaction time compared to RSA, with the main time reduction produced in the handshake.

    “Comparing ECDSA with RSA, it can be seen that, for 128 kilobyte payloads, ECDSA greatly reduces the total transaction time, being the main time reduction produced in the handshake as expected.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    With HTTP, 61 ACKs are sent, and with ECDSA, 46 ACKs are sent, resulting in an average of 0.63 ACKs per TCP frame in HTTP and 0.48 in ECDSA.

    “With HTTP, 61 ACKs are sent, and with ECDSA, only 46 are sent, which results in an average of 0.63 ACKs per TCP frame in HTTP and 0.48 ACKs per TCP frame in ECDSA.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingreasoning

    For 512-byte payloads using NGINX, the time differences between ECDSA and RSA are even larger than for 128 kilobyte payloads, because handshake savings remain constant and data transmission time is reduced.

    “For 512 byte payloads and using NGINX, the time differences are even larger, since the savings in the handshake remain constant, and the data transmission time is reduced.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingown data

    An RSA transaction takes approximately 0.012 s, while an ECC transaction takes 0.005 s.

    “When calculating such a metric, a result can be obtained that an RSA transaction takes approximately 0.012 s, while, when using ECC, it only takes 0.005 s.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    For HTTP with Apache, taking into account handshake and FIN, 190 frames should be sent if each TCP frame is ACKed, but only 160 frames are transmitted.

    “Taking into account the handshake (five frames) and FIN message (three frames), a total of 190 frames should be sent if, for each TCP frame, an ACK is also sent, but only 160 frames are transmitted.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingnone given

    By using more concurrent clients, the absolute energy consumption value varies, but the differences between implementations remain constant.

    “By using more concurrent clients, the absolute energy consumption value varies, but the differences between implementations remain constant.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingreasoning

    For a 128 kilobyte payload, RSA requires fewer frames than ECDSA, which requires fewer frames than HTTP.

    “Specifically, RSA is the one that requires less frames, followed by ECDSA and HTTP.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingown data

    The number of frames is almost the same in both cipher suites, but larger than without security due to TLS handshake messages transmitted at the beginning.

    “The number of frames is almost the same in both cipher suites, but such an amount is larger than without security, due to the TLS handshake messages transmitted at the beginning of the communications.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingown data

    Even with a 112-bit security level key for RSA and a more secure 128-bit security level key for ECDSA, ECC is more efficient at both the server- and client-side.

    “It is important to note that, in the experiments performed, a 112-bit security level key is being used for RSA in contrast with a more secure 128-bit security level key for ECDSA: even with this difference, ECC is more efficient at both the server- and the client-side.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    The total transmission time is higher when using GZIP, being more noticeable when using a 128 kilobyte payload.

    “The total transmission time is also higher, being more noticeable when using a 128 kilobyte payload.”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Findingown data

    The results with 16 concurrent clients are more stable than those obtained with less concurrency.

    “As it can be observed in the next figures, the results with this number of clients are more stable than the ones obtained with less concurrency.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingown data

    Apache presents very similar times, being even faster when RSA is used.

    “Apache presents very similar times, being even faster when RSA is used.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingown data

    For ECDSA with Apache, taking into account TLS handshake and FIN, 208 frames would be needed, but only 158 are transmitted.

    “Once again, taking into account the TLS handshake in this case (10 frames) and the FIN message (six frames), 208 frames would be needed, but only 158 are transmitted.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingown data

    For 128 kilobyte payloads, using no compression is always faster, with reductions in time of up to 46%.

    “For 128 kilobyte payloads, using no compression is always faster, with reductions in time of up to 46%.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingreasoning

    At the client side, ECDSA is still more energy-efficient than RSA, but the difference is less dramatic than at the server side.

    “As it can be observed, at the client-side, ECDSA is still more energy-efficient than RSA, but the difference is less dramatic than at the server-side.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingown data

    Using GZIP compression for a 512-byte payload presents some data reduction and heavily reduces data and number of frames for a 128 kilobyte payload.

    “Using GZIP compression for a 512 byte payload presents some data reduction and heavily reduces the data and number of frames transferred for a 128 kilobyte payload”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    For the security levels required nowadays, measurements indicate that ECC obtains power consumption reductions of up to 50% and a data throughput that doubles RSA in most scenarios.

    “Moreover, for the security levels required nowadays, measurements indicate that ECC obtains power consumption reductions of up to 50% and a data throughput that doubles RSA in most scenarios.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    In terms of throughput, ECC is always better than RSA, being able to double the number of requests per second when moderate levels of concurrency are reached.

    “In terms of throughput, ECC is always better than RSA, being able to double the number of requests per second when moderate levels of concurrency are reached.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingreasoning

    A detailed frame analysis showed that the handshake takes more time in the RSA cipher suite than in the ECC one, with remaining data transmission time unaffected.

    “A detailed frame analysis showed that the handshake takes more time in the RSA cipher suite than in the ECC one, with remaining data transmission time unaffected.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    For payloads of less than 128 bytes, the use of compression causes larger amounts of data to be transmitted, and for any of the payload sizes tested, slower transactions and higher power consumption.

    “The tests also demonstrated that in the case of transmitting payloads of less than 128 bytes, the use of compression causes larger amounts of data to be transmitted, and, for any of the payload sizes tested, slower transactions and higher power consumption.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    There is a small difference in power consumption and throughput between both servers, but a substantial difference between RSA and ECC across all tests performed.

    “As it is shown in the experiments, there is a small difference in power consumption and throughput between both servers, but a substantial difference can be observed between RSA and ECC across all the tests performed.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingreasoning

    The total data transmitted is almost the same between both cipher suites, although they differ in the number of frames exchanged due to cumulative ACKs.

    “Moreover, the total data transmitted is almost the same between both cipher suites, although they differ in the number of frames exchanged due to cumulative ACKs.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingown data

    ECC greatly outperformed RSA in terms of both power efficiency and communications throughput.

    “ECC greatly outperformed RSA in terms of both power efficiency and communications throughput.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    The server-side power consumption when using RSA doubles the one obtained with ECC, while the client-side presents a 15-20% lower power consumption for the ECC scheme.

    “The server-side power consumption when using RSA doubles the one obtained with ECC, while the client-side presents a 15–20% lower power consumption for the ECC scheme.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Premisecitation

    There is research on energy efficiency of IoT gateways, but it is usually assumed that they have direct access to the grid, which is not always the case.

    “There is also some research on the energy efficiency of the IoT gateways [7], but it is usually assumed that they have direct access to the grid, which is not always the case”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites [7] Saxena, 2017, Efficient IoT Gateway over 5G Wireless: A New Design with Prototype and Implementation Results, IEEE Commun. Mag. source

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation

    There are two main subgroups of cipher suites recommended for TLS: RSA and ECDHE.

    “There are two main subgroups of cipher suites recommended for TLS [31]: RSA and ECDHE.”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [31] Polk, T., McKay, K., and Chokhani, S. (2005). Guidelines for the Selection and Use of Transport Layer Security (TLS) Implementations, NIST. NIST Special Publication 800-52 Revision 1.

      Not checked in the free run.

  • May combine two claims.

    Premisereasoning

    The key sizes needed for RSA in widespread TLS cipher suites are not practical for most IoT devices and cannot be scaled to high security levels.

    “but the key sizes needed are not practical for most IoT devices and cannot be scaled to high security levels.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Premisecitation

    Security in IoT and fog computing systems is often overlooked or not completely addressed, slowing down the broad adoption of IoT.

    “security in IoT and fog computing systems is often overlooked or not completely addressed [18,21,22], slowing down the broad adoption of IoT.”
    Quote in the paperSubstantial claim0 of 3 sources checked
    • cites [18] Fernández-Caramés, T.M., Fraga-Lamas, P., Suárez-Albela, M., and Castedo, L. (2016). Reverse Engineering and Security Evaluation of Commercial Tags for RFID-Based IoT Applications. Sensors, 17. source

      Not checked in the free run.

    • cites [21] Asplund, 2016, Attitudes and perceptions of IoT security in critical societal services, IEEE Access source

      Not checked in the free run.

    • cites [22] Mukherjee, B., Neupane, R.L., and Calyam, P. (2017, January 26–28). End-to-End IoT Security Middleware for Cloud-Fog Communication. Proceedings of the 2017 IEEE 4th International Conference on Cyber Security and Cloud Computing (CSCloud), New York, NY, USA. source

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Interpretationreasoning

    The IoT fog layer exchanges payloads with a wide range of sizes.

    “This leads the IoT fog layer to exchange payloads with a wide range of sizes.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Premisecitation

    For a 112 security level, RSA requires a 2048-bit key while ECC requires a 224-bit key.

    “for a 112 security level, a 2048-bit key size must be used if an RSA cipher is selected, but only a 224-bit key size is needed if the cipher uses ECC.”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [73] Barker, E., Barker, W., Burr, W., Polk, W., and Smid, M. (2012). Recommendation for Key Management—Part 1: General (Revision 3), NIST. NIST Special Publication 800-57 Part 1, Revision 3. source

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation

    768-bit RSA was factored in 2010 using the number field sieve factoring method.

    “768-bit RSA was factored in 2010 using the number field sieve factoring method [69]”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [69] Kleinjung, 2010, Factorization of a 768-Bit {RSA} Modulus, Crypto

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

    Premisereasoning

    Requests exchanged between IoT nodes from the same network consist of small payloads, whereas transactions between different IoT networks are usually larger because protocol translations may be needed.

    “The requests exchanged between IoT nodes from the same network consist of small payloads, whereas transactions between different IoT networks are usually larger, since protocol translations may be needed.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation

    A 1024-bit RSA implementation of OpenSSL was successfully broken using a fault-based security attack in less than 100 h.

    “a 1024-bit RSA implementation of OpenSSL was successfully broken using a fault-based security attack in less than 100 h [70].”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [70] Pellegrini, A., Bertacco, V., and Austin, T. (2010, January 8–12). Fault-based attack of RSA authentication. Proceedings of the 2010 Design, Automation & Test in Europe Conference & Exhibition (DATE 2010), Dresden, Germany. source

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation

    Most IoT deployments are based on TCP/IP communications.

    “Most IoT deployments, such as the home automation system evaluated in [13], are based on TCP/IP communications.”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [13] Suárez-Albela, M., Fraga-Lamas, P., Fernández-Caramés, T.M., Dapena, A., and González-López, M. (2016). Home Automation System Based on Intelligent Transducer Enablers. Sensors, 16. source

      Not checked in the free run.

    • The authors state this as an assumption: IoT fog gateways support TCP/IP.

  • Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

    Premisecitation

    Certificates can also be used with CoAP, but the cipher suites supported have low security levels.

    “Certificates can also be used, but the cipher suites supported have low security levels.”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [60] (2017, July 28). Using Raw Public Keys in Transport Layer Security (TLS) and Datagram Transport Layer Security (DTLS). Available online: https://tools.ietf.org/html/rfc7250.

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisereasoning

    The use of PSK cipher suites constrains devices to a very specific scenario because PSK has to be established manually and set up in advance.

    “The use of this type of cipher suites constrains the devices to a very specific scenario (e.g., since PSK is used, it has to be established manually and set up in advance).”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisereasoning

    ECDHE-based cipher suites provide Perfect Forward Secrecy because the key is regenerated after each session, in contrast to variants based on RSA.

    “The Ephemeral part means that the key is regenerated after each session, providing Perfect Forward Secrecy (PFS) in contrast to the variants based on RSA.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based).

    Interpretationnone given

    The vast majority of IoT fog computing systems will follow the layered architecture presented in Figure 1 to some extent.

    “Nevertheless, the vast majority of IoT fog computing systems will follow this architecture to some extent.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based). May combine two claims.

    Interpretationreasoning

    It is not always possible to accommodate the use of TLS resumption on IoT fog gateways, and, depending on the number and periodicity of the connections, its benefits can be limited.

    “Moreover, it is not always possible to accommodate the use of TLS resumption on IoT fog gateways, and, depending on the number and periodicity of the connections, its benefits can be limited.”
    Quote in the paperNot a substantial claim
    • The authors state this as an assumption: IoT fog gateways support TCP/IP.

  • Classified as probably not a substantial claim. May combine two claims.

    Premisecitation

    IoT gateways and nodes are resource-constrained and usually embed low-cost computing devices that consume little power.

    “In comparison to the cloud or to regular computers, IoT gateways and nodes are resource-constrained and usually embed low-cost computing devices that consume little power [23].”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites [23] Sehgal, 2012, Management of resource constrained devices in the internet of things, IEEE Commun. Mag. source

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based).

    Premisenone given

    2048-bit RSA is the minimum size considered secure nowadays.

    “2048-bit RSA being the minimum size considered secure nowadays.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S4 (based). May combine two claims.

    Interpretationreasoning

    IoT gateways that need to manage a great number of connections are even more affected by encryption overhead, reducing throughput and increasing power consumption.

    “Moreover, IoT gateways, that need to manage a great number of connections are even more affected by this encryption overhead, reducing throughput, and increasing power consumption.”
    Quote in the paperNot a substantial claim
    • The authors state this as an assumption: IoT gateways are constrained in terms of computational resources.

  • Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based).

    Premisereasoning

    If the prime numbers used are big enough, the resultant RSA encryption is considered secure.

    “If the prime numbers used are big enough, the resultant encryption is considered secure.”
    Quote in the paperNot a substantial claim
    • The authors state this as an assumption: The RSA trapdoor function is based on the assumption that factorizing large integer numbers that are the result of two large prime numbers is a difficult problem to solve.

  • Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

    Premisecitation

    CoAP was designed having resource-constrained devices in mind, so security is addressed lightly, with limitations such as the use of pre-shared keys or raw public key cipher suites.

    “CoAP was designed having resource-constrained devices in mind, so security is addressed lightly, with a lot of limitations, such as the use of pre-shared keys or raw public key cipher suites [60].”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [60] (2017, July 28). Using Raw Public Keys in Transport Layer Security (TLS) and Datagram Transport Layer Security (DTLS). Available online: https://tools.ietf.org/html/rfc7250.

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation

    DHE is clearly outperformed by ECDHE.

    “Note that DHE can also be used, but it is clearly outperformed by ECDHE [64].”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [64] Huang, 2014, An Experimental Study of TLS Forward Secrecy Deployments, IEEE Internet Comput. source

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

    Premisecitation

    The PSK cipher suite PSK-AES-128-CCM-8 is considered insecure and not recommended.

    “On top of that, this cipher suite is also considered insecure and not recommended [31].”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [31] Polk, T., McKay, K., and Chokhani, S. (2005). Guidelines for the Selection and Use of Transport Layer Security (TLS) Implementations, NIST. NIST Special Publication 800-52 Revision 1.

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Interpretationreasoning

    The classification of constrained devices in [28] fails to give an updated definition of constrained device because of the fast evolution of embedded hardware capabilities.

    “Because of the fast evolution of embedded hardware capabilities, this classification fails to give an updated definition of constrained device.”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [28] (2017, July 28). Terminology for Constrained-Node Networks. Available online: https://tools.ietf.org/html/rfc7228.

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S4 (based).

    Interpretationnone given

    The use of a 2048-bit certificate on an ephemeral key-exchange algorithm introduces heavy overhead and computing requirements, which are very difficult to accommodate on the constrained hardware capabilities of most IoT devices.

    “However, note that the use of a 2048-bit certificate on an ephemeral key-exchange algorithm introduces heavy overhead and computing requirements, which are very difficult to accommodate on the constrained hardware capabilities of most IoT devices.”
    Quote in the paperNot a substantial claim
    • The authors state this as an assumption: IoT gateways are constrained in terms of computational resources.

  • Classified as probably not a substantial claim. May combine two claims.

    Study design and conductnone given

    The Adafruit INA219 offers enough precision, measuring 0.1 mA steps with a maximum of ±400 mA or 0.8 mA steps with a maximum of ±3.2 A, and allows measuring up to 26 V.

    “an Adafruit INA219 was selected, since it offers enough precision (it can be configured in high precision mode measuring 0.1 mA steps with a maximum of ±400 mA or in low precision mode measuring 0.8 mA steps and a maximum of ±3.2 A), and allows for measuring up to 26 V.”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim: may not pass S2 (checkable), S4 (based).

    Study design and conductnone given

    To provide medium- and long-term valid conclusions, the selected hardware must fulfill future performance expectations for energy-efficient IoT gateways.

    “Second, in order to provide medium- and long-term valid conclusions, the selected hardware must fulfill future performance expectations for energy-efficient IoT gateways.”
    Quote in the paperNot a substantial claim
    • Rests on an assumption: The selected hardware fulfills future performance expectations for energy-efficient IoT gateways.

  • Probably not a substantial claim: may not pass S4 (based). May combine two claims.

    Interpretationreasoning

    Using a different SBC or SoC-based platform could yield different total power consumption results, but the conclusions obtained when comparing cipher suites should not change.

    “Using a different SBC or SoC-based platform could yield different total power consumption results, but the conclusions obtained when comparing cipher suites should not change.”
    Quote in the paperNot a substantial claim

  • Classified as probably not a substantial claim. May combine two claims.

    Study design and conductnone given

    The Allwinner H3 SoC's hardware-accelerated encryption support was not used during the tests because it can distort the results obtained.

    “Although the Allwinner H3 SoC has hardware-accelerated encryption support powered by a Crypto Engine (CE), it was not used during the tests, since it can distort the results obtained.”
    Quote in the paperSubstantial claim

  • Classified as probably not a substantial claim.

    Study design and conductnone given

    Both Apache2 and Nginx used the same TLS library, OpenSSL.

    “In addition, both servers used the same TLS library (OpenSSL).”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Premisereasoning

    Existing works comparing ECC and RSA power consumption and performance have drawbacks such as not accounting for security levels, using ad hoc implementations, comparing in isolation, providing approximated energy consumption, not measuring with external devices, using hardware platforms that do not provide mid-term valid results, not using real communication protocols, or using insecure key lengt

    “Not taking into account security levels, comparing the algorithms with an ad hoc implementation and in an isolated way, providing an approximated value of the energy consumption or not measuring it with external devices, or the use of hardware platforms that do not provide mid-term valid results are some of the drawbacks found.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Interpretationreasoning

    The increment in energy consumption with GZIP for large files is due to compression taking more time and power, and the gains of sending less data not compensating in total energy consumption.

    “This is due to the fact that the compression of large files takes more time and power, and in a general IoT scenario, the gains of sending less data through the network do not compensate for the effort of compressing data in terms of total energy consumption.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Premisecitation

    Both selected cipher suites are among the few recommended by NIST guidelines for cipher suite selection and configuration, sharing all parameters but the signing algorithm.

    “Both cipher suites chosen are among the few recommended by the National Institute of Standards and Technology (NIST) guidelines for cipher suite selection and configuration [31], sharing all the parameters but the signing algorithm.”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [31] Polk, T., McKay, K., and Chokhani, S. (2005). Guidelines for the Selection and Use of Transport Layer Security (TLS) Implementations, NIST. NIST Special Publication 800-52 Revision 1.

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Study design and conductreasoning

    The same tests were run in two different web servers to eliminate any bias that could be introduced by the server implementation or configuration.

    “During the experiments, the same tests were run in two different web servers to eliminate any bias that could be introduced by the server implementation or configuration.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Premisecitation

    The Cortex-A7 was chosen because it is the most efficient ARMv7-A processor and the most commercially successful with more than a billion units in production.

    “The Cortex-A7 [87] was chosen since it fits the requirements and it is the most efficient ARMv7-A processor and the most commercially successful with more than a billion units in production.”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [87] (2017, July 28). Cortex-A Series Family. Available online: https://www.arm.com/products/processors/cortex-a/cortex-a7.php.

      Not checked in the free run.

  • May combine two claims.

    Interpretationreasoning

    Compared to plain HTTP, the use of TLS is more CPU intensive and leads to longer lasting transactions that keep the CPU from idling.

    “Compared to plain HTTP, the use of TLS is more CPU intensive and leads to longer lasting transactions that keep the CPU from idling.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Study design and conductreasoning

    Using external hardware instead of the SBCs' ACPI to measure power consumption makes current readings independent from the particular ACPI-related hardware of the SBC and keeps the setup valid for any other SBC.

    “In this way, current readings are independent from the particular ACPI-related hardware of the SBC and the setup remains valid for any other SBC.”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Study design and conductnone given

    It was verified that both servers were configured for using all four logical processors available on the Orange Pi PC SOCs.

    “It was verified that both servers were configured for using all four logical processors available on the Orange Pi PC SOCs.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

    Study design and conductreasoning

    The time instants presented in the frame analysis figures are not the exact times when the data were received at the Orange Pi PCs, but the instants when the frames reached the desktop computer running Wireshark, and these times can vary and have to be treated as approximations.

    “Note that the time instants presented in the next figures are not the exact times when the data were received at the Orange Pi PCs, but the instants when the frames reached the desktop computer running Wireshark. While using a dedicated switch and a controlled network environment, these times can vary and have to be treated as approximations.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Interpretationreasoning

    In the interval when TLS processes a received frame, other frames can arrive, and only the ACK is sent to confirm all of them, reducing the number of ACKs and total data transmitted.

    “In this interval of time, other frames can arrive, and, in this case, only the ACK is sent to confirm all of them. Thus, the number of ACKs and the total data transmitted are reduced.”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim: may not pass S2 (checkable), S3 (load-bearing). May combine two claims.

    Interpretationown data

    The throughput figure is a good example of how the implementation and configuration influence the servers and justifies the presence in the comparison of Apache and Nginx instead of showing results for only one server.

    “This figure is a good example of how the implementation and configuration influence the servers and justifies the presence in the comparison of Apache and Nginx instead of showing results for only one server.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based). May combine two claims.

    Interpretationnone given

    The delay in the Client Key-Exchange frame can be attributed to an Apache implementation or concurrency configuration performance drawback, although the specific cause remains unknown.

    “The reason for this delay can be attributed to some kind of performance drawback due to Apache implementation or concurrency configuration, although the specific cause remains unknown.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Interpretationreasoning

    Some references use outdated algorithms, insecure key sizes, or make unfair comparisons without taking into account security levels.

    “Some references use outdated algorithms, insecure key sizes, or make unfair comparisons without taking into account security levels.”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim: may not pass S2 (checkable), S3 (load-bearing), S4 (based).

    Interpretationnone given

    Other authors employ ad hoc implementations that are not useful in real IoT deployments.

    “Other authors employ ad hoc implementations that are not useful in real IoT deployments.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Interpretationcitation

    The client-side difference is due to RSA signing and verifying processes requiring asymmetric computational operation costs, while they are almost symmetric for ECDSA.

    “This difference is due to the fact that, in the case of RSA, signing and verifying processes require asymmetric computational operations costs, but they are almost symmetric for ECDSA, as other authors have also observed [84].”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites [84] Potlapally, 2006, A study of the energy consumption characteristics of cryptographic algorithms and security protocols, IEEE Trans. Mob. Comput. source

      Not checked in the free run.

  • May combine two claims.

    Interpretationown data

    The delay in the Client Key-Exchange frame explains the similar time values for ECDSA and RSA in Figure 23 and the gap in requests per second and requests per mWh between Apache and Nginx.

    “This explains why Figure 23 presents very similar time values for ECDSA and RSA, compared with Figure 22, where a clear difference can be seen. It also explains the gap in requests per second between Apache and Nginx that can be seen in Figure 14 and, consequently, the same gap in terms of requests per mWh in Figure 15.”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Study design and conductnone given

    Only server authentication was used in the tests.

    “It is also important to remark that, in the tests performed, only server authentication was used.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Interpretationown data

    Various references do not present energy consumption metrics, or their power consumption values are estimated and not measured with precise hardware.

    “Various references do not present energy consumption metrics, or the power consumption values are estimated and not measured with precise hardware.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Study design and conductnone given

    The chosen security library is OpenSSL, an open and widely used implementation of the TLS standard.

    “The chosen security library is OpenSSL, an open and widely used implementation of the TLS standard.”
    Quote in the paperNot a substantial claim

  • Classified as probably not a substantial claim.

    Study design and conductnone given

    Energy measurements are performed using isolated and precise hardware with a dedicated SBC.

    “On top of that, energy measurements are performed using isolated and precise hardware with a dedicated SBC.”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based). May combine two claims.

    Interpretationnone given

    By using ECC, the new emerging hardware platforms for end devices could overcome these limitations and perform as servers or implement client authentication.

    “By using ECC, the new emerging hardware platforms for end devices could overcome these limitations and perform as servers or implement client authentication.”
    Quote in the paperNot a substantial claim
    • Rests on an assumption: IoT end devices are usually treated as clients because the embedded hardware used to implement them cannot accommodate the necessary certificates and ciphering algorithms while maintaining reduced power consumption and reasonable response times.

  • Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based).

    Premisereasoning

    In an IoT fog scenario, gateways can act both as servers and clients since data requests can be horizontal and vertical, involving IoT nodes or other fog devices.

    “In an IoT fog scenario as the one described in Figure 1, the gateways can act both as servers and clients since data requests can be horizontal and vertical, involving IoT nodes or other fog devices.”
    Quote in the paperNot a substantial claim

  • Classified as probably not a substantial claim.

    Study design and conductown data

    The tests performed allowed for accurately determining the impact of each cipher suite on both sides of the communications.

    “The tests performed allowed for accurately determining the impact of each cipher suite on both sides of the communications.”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based).

    Study design and conductnone given

    The cipher suites and key sizes employed guarantee acceptable security levels for current and future IoT gateway developments.

    “Furthermore, the cipher suites and the key sizes employed guarantee acceptable security levels for current and future IoT gateway developments.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Study design and conductnone given

    In the tests performed only the server was using a certificate (only Server Authentication was used).

    “It is important to note that in the tests performed only the server was using a certificate (i.e., only Server Authentication was used)”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Interpretationreasoning

    Considering the near-term prospects of more secure levels, the key sizes needed in RSA will make it impractical not only for IoT fog computing applications, but for any secure connection.

    “Furthermore, considering the near-term prospects of more secure levels, the key sizes needed in RSA will make it impractical not only for IoT fog computing applications, but for any secure connection.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based).

    Study design and conductnone given

    The hardware testbed was chosen to remain comparable with future platforms in the years to come.

    “The hardware testbed was chosen to remain comparable with future platforms in the years to come.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Premisereasoning

    Both cipher suites only differ in the public key algorithms, the symmetric key and hash algorithms being exactly the same.

    “This was the expected result, since both cipher suites only differ in the public key algorithms, the symmetric key and hash algorithms being exactly the same.”
    Quote in the paperSubstantial claim

  • Classified as probably not a substantial claim.

    Study design and conductreasoning

    The tests presented in this paper take into account all of these details.

    “Contrarily, the tests presented in this paper take into account all of these details.”
    Quote in the paperSubstantial claim
91 of 91

  • The authors state this as an assumption.

    IoT gateways are constrained in terms of computational resources.

    “which are assumed to be constrained in terms of computational resources”

  • The authors state this as an assumption.

    IoT fog gateways support TCP/IP.

    “IoT fog gateways are assumed to support Transmission Control Protocol/Internet Protocol (TCP/IP).”

  • The authors state this as an assumption.

    The RSA trapdoor function is based on the assumption that factorizing large integer numbers that are the result of two large prime numbers is a difficult problem to solve.

    “the RSA trapdoor function is based on the assumption that factorizing large integer numbers that are the result of two large prime numbers is a difficult problem to solve.”

  • The paper relies on this without support.

    The impact of the signing algorithm used in the TLS handshake (ECDSA or RSA) is independent of the size of the payload transmitted by the underlying HTTP communications.

    “The impact of the signing algorithm used in the TLS handshake (i.e., ECDSA or RSA for the cipher suites selected) is independent of the size of the payload transmitted by the underlying HTTP communications.”

  • The paper relies on this without support.

    Typical payloads transmitted by IoT gateways when managing requests from a few IoT nodes are represented by the lower values used in these tests, from 32 to 1024 bytes.

    “Typical payloads transmitted by IoT gateways when managing requests from a few IoT nodes are represented with the lower values used in these tests (i.e., from 32 to 1024 bytes).”

  • The paper relies on this without support.

    IoT gateways can transmit larger payloads as a result of carrying out data aggregation for several IoT nodes.

    “Nevertheless, IoT gateways can transmit larger payloads as a result of carrying out data aggregation for several IoT nodes.”

  • The paper relies on this without support.

    GZIP is asymmetric, with the computational cost of compression greater than the computational cost of decompression.

    “These differences were expected, since GZIP is asymmetric, the computational cost of compression being greater than the computational cost of decompression.”

  • The paper relies on this without support.

    HTTP is the most used protocol on network communications.

    “HTTP is the most used protocol on network communications, hence we consider it as the best option in terms of consistency and implementation efficiency.”

  • The paper relies on this without support.

    The selected hardware fulfills future performance expectations for energy-efficient IoT gateways.

    “Second, in order to provide medium- and long-term valid conclusions, the selected hardware must fulfill future performance expectations for energy-efficient IoT gateways.”

  • The paper relies on this without support.

    Apache and Nginx are the two most extensively used web servers.

    “the two most extensively used web servers (Apache and Nginx) are employed and compared”

  • The paper relies on this without support.

    In most IoT scenarios, client authentication would also be needed.

    “but in most IoT scenarios, client authentication would be also needed.”

  • The paper relies on this without support.

    IoT end devices are usually treated as clients because the embedded hardware used to implement them cannot accommodate the necessary certificates and ciphering algorithms while maintaining reduced power consumption and reasonable response times.

    “IoT end devices are usually treated as clients, since the embedded hardware used to implement them is not able to accommodate the necessary certificates and the ciphering algorithms, while maintaining reduced power consumption and reasonable response times.”
12 of 12

Claims other papers make about this one

Each is checked against this paper’s own text: does it say what the citing paper says it does?

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

Each contradiction between two passages listed here was confirmed by an independent check.

  • Conclusion claims compression always raises power consumption

    Minor

    The conclusion says compression increased power consumption for every payload size tested, but the results explicitly report lower client-side consumption with compression for a 256-byte payload.

  • Conclusion claims compression always slows transactions

    Minor

    The conclusion says compression produced slower transactions for every tested payload size, while the frame analysis says Apache was faster with compression when RSA was used for a 512-byte payload.

2 of 2

Cites

Flagged first, then by citations

Research works printed but not linked: 2

  1. (2017, July 28). WolfSSL Manual—Chapter 4 (Features). Available online: https://www.wolfssl.com/wolfSSL/Docs-wolfssl-manual-4-features.html.
  2. Briggs, M.E. (1998). An Introduction to the General Number Field Sieve. [Master’s Thesis, Virginia Polytechnic Institute and State University].

Cited by

Sorted by citations

Related

Authors
Suárez-Albela, M., Fernández-Caramés, T., Fraga-Lamas, P., & Castedo, L.
Affiliations
  • Universidade da Coruña (ES)
  • Department Computer Engineering, Faculty of Computer Science, Universidade da Coruña, 15071 A Coruña, Spain
Corresponding author
Suárez-Albela, Manuel
ORCID
Venue
Sensors
Publisher
MDPI AG
Published
2017-08-29
Volume
17
Issue
9
Pages
1978
ISSN
1424-8220 (ISSN-L)
DOI
10.3390/s17091978
PMC
PMC5620735
OpenAlex
W2752575750
Semantic Scholar
5c258e7ed26da9b4f2947e4c056471bf68961e58
Cited by
125
References
79
Citations by year
2026: 3 · 2025: 7 · 2024: 6 · 2023: 12 · 2022: 16 · 2021: 18 · 2020: 11 · 2019: 26 · 2018: 25 · 2017: 1
FWCI
11.43
Type
journal article
Language
en
Open access
Licence
Locations
  • Sensors (publishedVersion, cc-by)
  • PubMed Central (publishedVersion, cc-by)
  • PubMed (publishedVersion)
  • LA Referencia (Red Federada de Repositorios Institucionales de Publicaciones Científicas) (submittedVersion, cc-by)
  • DOAJ (DOAJ: Directory of Open Access Journals) (submittedVersion, cc-by-sa)
  • RUC (Universidade Da Coruña) (submittedVersion, cc-by)
Funders
  • Xunta de Galicia (ED431C 2016-045, ED341D R2016/012, ED431G/01)
  • Agencia Estatal de Investigación of Spain (TEC2013-47141-C4-1-R, TEC2015-69648-REDC, TEC2016-75067-C4-1-R)
  • ERDF funds of the EU (AEI/FEDER, UE)
Topics
  • Physical Sciences › Computer Science › Computer Networks and Communications › IoT and Edge/Fog Computing
  • Physical Sciences › Computer Science › Artificial Intelligence › Cryptographic Implementations and Security
  • Physical Sciences › Computer Science › Computer Networks and Communications › Advanced Authentication Protocols Security
Keywords
IoT gateway, fog computing, energy efficiency, power consumption, elliptic curve cryptography, public key exchange, data throughput, data compression, resource-constrained devices, battery-powered devices, Internet of Things security

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

Claims read by deepseek-v4.1-flash under definition version 3 (extract-v3) on 2026-10-09.

Classified by jev-1.13-20260917 (gold-guideline-v0.2) on 2026-10-09.

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review