A survey on Iot security: Application Areas, Security Threats, and Solution Architectures
tải về 6.16 Mb. Chế độ xem pdf
|
thong-tin-so bao-cao-qpsk-1 - [cuuduongthancong.com]
- Điều hướng trang này:
- IX. OPEN ISSUES, CHALLENGES, AND FUTURE RESEARCH DIRECTIONS
| B. CHALLENGES IN EDGE LAYER
Although edge computing provides various features to in- crease the security and performance of IoT applications, there are various challenges associate with completely relying on the edge layer for all computation. Edge devices include sen- sors, RFID devices, actuators, tags, and embedded devices. The edge layer is highly susceptible to attacks in an IoT system. If the edge layer is compromised, then the entire system may be compromised. MQTT and COAP are the most popular protocols for the edge layer. Both these protocols do not use any security layer by default. Although the option to add an optional security layer in the form of TLS for MQTT and DTLS for COAP is present, it creates additional overhead in terms of processing and bandwidth. Issues specific to edge devices include sleep deprivation attacks, battery draining at- tacks, and outage attacks. Edge devices are typically resource constrained, and the most important resource they rely upon is the battery backup. The foremost and easiest way to attack the edge devices is to somehow deplete the battery of an edge device. For example, an attacker might force the edge device to do some power hungry or infinite loop computation [197]. The process of striking a balance between storing and processing data on edge or cloud is very important. Keeping too much data on edge may also lead to overwhelming of the edge devices and may impact the entire application. IX. OPEN ISSUES, CHALLENGES, AND FUTURE RESEARCH DIRECTIONS There are some performance and security issues in the use of blockchain, fog computing, edge computing and machine learning for IoT security that are yet to be solved. This section discusses some of these issues. The security of blockchain depends on its method of implementation and the use of software and hardware in that implementation. Since all the transactions made by users in blockchain are public, there is a possibility that private information of users can be revealed. Also, as the number of miners increases, the size of blockchain also increases continuously. This increases the cost of storage and reduces the speed of distribution over the whole network, giving rise to issues like scalability and availability of blockchain [198]. Since fog computing is a nontrivial extension of cloud computing, some of the issues such as security and privacy will continue to persist [120]. Therefore, before implement- ing fog-assisted IoT applications, these security and privacy goals of fog computing are required to be studied. Some of the challenges and research issues on security and privacy in IoT environments and the solutions provided by fog comput- ing are discussed in [127]. There are many machine learning algorithms in existence. Therefore, it is imperative to select a proper algorithm suit- able for the application. Selecting a wrong algorithm would result in producing “garbage” output and will lead to loss of effort, effectiveness and accuracy. Similarly, choosing the wrong data set will lead to “garbage” input producing incorrect results. The success of a machine learning solution depends on these factors as well as diversity in selecting data. If the data is not clustered and classified, the prediction accu- racy will be lower. Also, the historical data may contain many ambiguous values, outliers, missing values, and meaningless data. IoT applications are creating a huge amount of data, and therefore it is a difficult task to clean and preprocess that data accurately. Various features like attribute creation, linear regression, multiple regression, removing redundancies and compressing data are required to effectively use machine learning for securing the IoT. In case of edge computing, data security and user privacy are the main concerns. An user’s private data can be leaked and misused if a house that is deployed with IoT devices is subjected to cyber attacks. For example, a person’s presence or absence at home can be revealed simply by observing the electricity or water usage data. Since the data is computed at the edge of data resource (e.g., home), therefore, the user has to be aware of some of the measures like securing WiFi connections. Secondly, data at edge should be owned fully by the user, and he/she should have control on which data to be shared. Some of the future research directions in this field are: • The edge devices are most resource constrained devices in the IoT and are therefore uniquely vulnerable to attacks. Penetration studies show that while it takes very little power to implement best practice security for the edge nodes, they are still highly vulnerable to a variety of malicious attacks. • The gateways between different layers in the IoT system need to be secured. Gateways provide an easy entry point for the attackers into the IoT system. End to end encryption, rather than specific encryption techniques for specific protocols would be a promising solution to secure the data passing through the gateways. The data should be decrypted only at the intended destination and not at the gateways for protocol translation. • Inter-fog sharing of resources is one of the areas where further work needs to be done. As of now, when the fog layer is unable to process the requests due to heavy load, the requests are forwarded to the cloud. There can be resource sharing between neighboring fog layers to prevent unwanted requests to be transferred to the cloud. • The current blockchain architecture is highly limited in terms of the number of nodes in permissioned networks 18 VOLUME x, 2019 This work is licensed under a Creative Commons Attribution 3.0 License. For more information, see http://creativecommons.org/licenses/by/3.0/. This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/ACCESS.2019.2924045, IEEE Access Vikas Hassija et al.: A Survey on IoT Security: Application Areas, Security Threats, and Solution Architectures and in terms of throughput in permissionless networks. Various consensus algorithms are being designed to support high throughput along with a large number of nodes or users. • Fog layer can be made more intelligent using various ML and AI techniques. Fog layer must be able to decide the duration for which the data in the fog should be retained and when the data should be discarded or shifted to the cloud for prolonged storage. • More efficient and reliable consensus mechanisms can be designed to reach consensus among the nodes along with preventing rampant use of computation power. The current consensus algorithms are highly resource hungry and less efficient. • The tamper-proof feature of blockchain is ending up into a collection of a lot of garbage data and addresses. There is a lot of invalid data that is never deleted like the addresses of the destructed smart contracts. This affects the performance of the overall application and better ways need to be designed to efficiently handle the garbage data in the blockchain. • Data analysis in near real-time and in the proximity of the IoT node is crucial for successful deployment of IoT applications. Various ML-based algorithms can be designed to analyze the data in the node itself to prevent the data transit for analysis. This can further enhance the security of the application by preventing data movement. tải về 6.16 Mb. Chia sẻ với bạn bè của bạn:
|