Data created by Internet of Things (IoT) sensors must be secured better, say some. A simple password-on-device solution is no longer sufficient thanks to increasing data protection regulations, a new public awareness of tracking, and hugely proliferating devices.\u00a0\nA new kind of architecture using Security Agents should be aggressively built into local routers and networks to handle IoT security and computation rather than offloading the number-crunching to a data center or the cloud, or indeed trying to perform it on the resource-limited IoT device, IEEE researchers say. In other words, IoT security should be handled at the network level rather than device for best results.\nBoxes containing the necessary hardware could be developed specifically for the purpose, where traditional, but newly security-modified, routers aren\u2019t applicable\u2014such as in a smart vehicular setting, like by the side of a road.\n\nNew IoT security framework\nA \u201cnew security framework is required,\u201d say the authors of the paper, Reconfigurable Security: Edge Computing-based Framework for IoT\u00a0(pdf), published by IEEE Network.\nThose researchers, consisting of IEEE members and a fellow, say existing IoT security protections, which they explain generally consist of just an authenticated key exchange and access control for communications, must be upped.\n\u201cAnonymous protection and fine-grained secure access control\u201d has to be included in the solution, they say. That includes techniques for guarding against malicious IoT devices, too.\nThe problem is small IoT sensors and radios can\u2019t handle the cryptographic processing required for truly diverse, customizable security, like this now-required anonymization. And as people become more conscious of the need to secure their data, they are going to want that data to remain private. The \u201csensitive behaviors of each individual\u201d need protecting against location traceability, is how the IEEE team put it.\n\u201cA new reconfigurable security framework for IoT (ReSIoT),\u201d based on edge computing will fix the problem, the researchers say. Their solution: a new, IoT management component to be installed at the edge nearby\u2014the Security Agent.\n\n\n\n\n\nRouters, base stations, and other near-edge boxes acting in this new security role would handle the computing that the IoT device can\u2019t (due to size, power limitations, and so on). That includes the intensive cryptographic stuff. The researchers say this will not only be more secure, but it will also simplify the management of keys\u2014cryptographic key disclosure risk increases as more keys, or passwords, need to be implemented by applications. The solution would also be more scalable.\nThat the IoT device doesn\u2019t need upgrading periodically is an added benefit. In fact, all you need to do is manage the Security Agent box, which can operate numerous, possibly remotely located sensors that are difficult to get to. IoT device cost, too, would be kept in check, as sensors can remain inexpensive\u2014no need to modify them and add to their cost to handle security.\n\u201cEven the low-end devices will be able to be protected by advanced security algorithms requiring high computation costs,\u201d the researchers explain.\nIoT processing time over legacy systems could, overall, start increasing because of security, the researchers say. Using Security Agents is better partly because the \u201chigh complexity computations\u201d are offloaded to the Security Agent, which has more power. The \u201cprotocols outperform legacy solutions,\u201d they claim. \u201cEven though additional communications [are] required to interact with [Security Agents].\u201d\nThere would also, presumably, be some edge-gained distance savings in comparison to an in-cloud key management solution.\nInterestingly, the group points out that IoT applications will fail if the now-needed and very intensive authentication can\u2019t be completed fast enough. IoT eats itself, in that case, making it clear something has to be done.