Understanding the trends in Intelligent Multi-Modal Security Systems (IMSS) will enable manufacturers, consultants, and system integrators to anticipate the requirements at the time of deployment. For system operators, knowing the trends keeps systems current to reduce risk. Here are several key trends to monitor.

The Evolution of Intelligent Multimodal Security Systems

As we progress from distributed systems with distributed machine learning with end-to-end security and enter the age of predictive analytics, augmented humanity,Footnote 1 telepresence, mirrorworlds,Footnote 2 and transhumanism,Footnote 3 security capabilities must anticipate the confidentiality, privacy, integrity, trustworthiness, and availability requirements for IMSS.

Intelligence at the edge

Fundamental economics and response speed requirements are driving intelligence to edge devices. Not only does that mean that Edge devices are more expensive due to the cost of Machine Learning (ML) applications and cost of specialized hardware to make them efficient enough to be useful but also increases edge device value in terms of risk because of reliance on the intelligent edge devices. The devices are more valuable because the loss has a bigger impact on the overall security objective.

More intelligence on the edge means not only more need for cybersecurity but also more need for ML-specific security. As ML-fueled machine-on-machine attacks evolve, defense must also stay in front so the benefits of the investment in intelligent edge devices can be realized and be beneficial to mankind.

Chapter 5 informed us how machine learning is rapidly evolving not just in algorithm development for utilitarian use cases, but also evolving as a threat, and as defenses against classic cybersecurity threats as well as threats specific to machine learning.

Multimodal

Emerging IMSS that perform machine learning on multiple sensors to extract a full situational awareness are also subject to errors and manipulation from each of these inputs. Sensor fusion must account for the security, accuracy, and resilience of each input and apply more confidence to the more resilient inputs and apply increased skepticism to the less resilient inputs.

Mobility

Edge devices for IMSS are going mobile. Dashboard cameras, body cameras, and drones are all not only pushing the boundaries of where IMSS can reach but also exposing the systems to more threats because of the lack of physical protections for these devices. The energy constraints of battery technology place stress on the budget for cybersecurity. Yet, coupled with technologies like GPS, Wi-Fi communications, and location services, and cellular communications with e-911 services, it also means more information can be exposed if the systems are not properly secured.

Threats

In Chapter 4, the evolution of threats was discussed. Attackonomics will provide incentive for this for the foreseeable future, and malware, tailored malware, targeted threats, weaponization of research, and machine learning will continue to evolve. New and valuable assets as components of or protected by IMSS will also drive this trend. Most of today’s encryption is based on the difficulty of solving certain classes of math problems, for which brute force solutions take years to complete. Quantum computers can break today’s encryption methods in seconds, not years. Quantum computers are still in research today, but are expected to be able to break today’s encryption by 2030.

Defenses

Defenses are also evolving. Defense in depth, quantum resistant cryptography, advances in privacy preservation, robustness, transparency and ethics in machine learning, machine learning performing specific defensive functions, and zero trust methods are all helping to solve problems making IMSS increasingly secure, robust, and resilient.

Trust

We are in an age of a bimodal divergence of trust. Trust in technology, government, news media, business, the 1%, the other 99%, and even the integrity of documentary evidence has been deteriorating for many years. The emergence and proliferation of what manifests as insider attacks (usually remote attackers aided by phishing, malware with privilege escalation) also erodes the underpinning assumptions of trust in colleagues, networks, and devices in the enterprise.

Privacy

Another value proposition of edge device intelligence is improvements in privacy protection. Take the case where a security system camera is detecting and identifying people. Applying security principles, the output video feed should be redacted to preserve identities and unredacted identifying information transmitted with encryption in a side band data channel so authorized parties with permission can access the identifying information. This provides improved system privacy protection over systems that send unredacted video everywhere (even if it is encrypted in communication links) by making sure that unredacted video is never available except when in a controlled authorized environment. However, if the security of the camera is poor, a remote hacker can turn off the redaction and the privacy benefit is lost. Once again, value in IMSS edge devices depends on corresponding security for the benefit to be realized.

It is common for technology to run ahead of regulatory and standards activity. Certainly, the open market will arrive at solutions that would never be invented by regulating them into existence, though that same open market inadvertently enables harm from unethical use and unforeseen consequences. Regulations and Standards, and even new technologies respond to ameliorate negative consequences, but the response takes time.

The growing body of privacy and data sovereignty legislation from municipalities, states, nations, and federations of nations requires not only rigor in system design and integration, but also continued vigilance. Fielded systems may have architectures or capabilities that must be modified or retracted in the future. And data laws can require data to be moved or removed entirely from systems.

National and International standards that apply to IMSS and Machine Learning applications are primitive today.

Given all this change, what should you be doing?

What Should You Do?

If you are reading this, it is out of date.

The speed at which technology evolves, and laws, regulation, and standards emerge means that Component and Software suppliers, IMSS Equipment Manufacturers, systems integrators, consultants, and system operators must actively monitor progress to properly maintain their systems.

So, pay attention! Manually searching out all this information would be a full time job for a good sized team. Fortunately, today there are many information and news aggregator services that feature relevant cybersecurity, legal, policy, and standards threads.Footnote 4, Footnote 5, Footnote 6 The Washington Post issues daily newslettersFootnote 7 on many relevant topics. On standards and recommendations, ANSI,Footnote 8 ISO,Footnote 9 IEC,Footnote 10 and NISTFootnote 11 provide regular newsletters. Start your day with a cup of your favorite morning beverage and a browse of cybersecurity news.

Laws, Regulations, and Public Policy are mandatory for IMSS operators to comply with to avoid liability and penalties. IMSS legislation comes from topical areas of physical security, cybersecurity, information security, privacy, and specific topics around the use of Computer Vision, Artificial Intelligence, and Machine Learning. Laws and legal precedent from judgments lag technology. So, while tracking these is necessary, it is not sufficient to manage risk. Trade associations are helpful, not only to stay informed, but also trade associations monitor and actively participate in the creation of laws and public policy to improve their applicability and quality. ASIS International,Footnote 12 the Security Industry Association (SIA),Footnote 13 and the PSAFootnote 14 are physical security trade associations that publish information on Laws and Regulations.

Laws and policies are mostly written by lawyers, not engineers, and they take long enough to get passed that technological details would not be applicable by the time they are enacted. Consequently, Manufacturers and System Integrators need to understand and monitor progress in technical standards. IMSS are members of the family of Internet of Things (IoT) devices. The leading relevant IoT standards and recommendations are:

  • NISTIR 8259Footnote 15 Foundational Cybersecurity Activities for IoT Device Manufacturers

  • ISO/IEC 27402Footnote 16 Cybersecurity – IoT security and privacy – Device baseline requirements

  • ETSI 303 645Footnote 17 Cybersecurity for Consumer Internet of Things: Baseline Requirements

  • CTA 2088Footnote 18 Baseline Cybersecurity Standard for Devices and Device Systems

Your system started its life with appropriate security. To keep it secure, you must make sure you are able to efficiently maintain your system. First, make sure your suppliers can contact you with notifications and updates. Some OSs and applications have an automatic update feature that is helpful. Even some OEMs will have an update service for the SW stack and applications sourced from them. Your firewall, anti-malware, and anomaly monitoring tools should also automatically update themselves. Also, some security maintenance applications will automate this for you. Device AuthorityFootnote 19 and RazberiFootnote 20 supply good examples of complete lifecycle management tools, from onboarding onward. And, don’t neglect your hardware – drivers are SW components, but are generally associated with the hardware components, so make sure they are included in your maintenance portfolio.

For all of these, (securely!) use the computerness of your computer and the Internetness of the Internet, and leverage applications created by experts in cybersecurity to accomplish these essential tasks with the best speed and the minimum amount of manual effort.