MCollections——16

MCollections——16

Design of Multipath Transmission Control for Information-Centric Internet of Things:

A Distributed Stochastic Optimization Framework

以信息为中心的物联网多路径传输控制设计:一个分布式随机优化框架                                                           

Crowds’ increasingly more intelligent devices are now capable of sensing, computation, and communication, which produces opportunities for more novel applications, such as video surveillance,e-health, and smart home/city. These emerging applications facilitate the intelligence of our daily life, and network changes its main function from device communication to data sharing. The traditional D2D paradigm has to map data to its carrier and establish end-to-end links between each of requester-provider pairs and is high demand yet significantly unutilized of precious communication resources.

Information-centric networking(ICN) has the nature of fast dissemination of massive data pushed up by increasing loT devices. To handle the massive data and overcome the limitations posed by performing hop-by-hop on globally optimizing the control, as

intermediate nodes determining the sending rate by the link condition within one-hop range might not be well understanding the available capacity of the entire path, redundancy and multipath transport can also be used to improve reliability in other contexts, such as the Internet of Things(loT). However, ICN’s data naming scheme and inherent multipath delivery hasn’t been up to its potential performance since providing optimal multipath-oriented transmission control is crucial for ICN-loT data delivery which is challenging.Because of the random movement of loT devices, dynamic link condition, and on-path caching, the optimal multipath-oriented transmission control cannot be handled using deterministic methods and concern is raised that revisiting the transmission control for achieving desired data rate as well as providing reliable and timely delivery.More prominently, the resource limitation and scalability issues in loT add more obstacle in its design. Consequently, lightweight and distributed control scheme is required.

To control the information-centric data transmission, the simplest way is to modify the existing IP-based transport control mechanism to adopt the ICN receiver-driven design.In this paper, the authors propose a distributed stochastic optimization framework for multipath transmission control in ICN-loT including request scheduling and data rate regulation, which is formulated as a stochastic concave optimization problem. The constructed problem aims to accommodate the randomness, unpredictability, and multipath delivery of ICN-loT and maximize the overall throughput, and can be linearly separated into two subproblems:1)a request scheduling problem ,and 2) a data rate control problem, which can be individually solved per time slot.A distributed alternating descent method(DADM) is designed to optimally control the transmission by solving the aforementioned problems at client sides. When implementing the DADM algorithm, each client sequentially update the request schedule and rate regulation via communicating the links and providers they use. Theoretical results show that DADM asymptotically converges to optimality while allowing low-complexity and decentralized implementation. Validated by simulations, DADM significantly improves throughput, delay reduction, and energy efficiency, in comparison with other state-of-the-art solutions.

The contributions of this paper are as follows:

1. The authors characterize the transmission control problem for ICN-loT as a stochastic concave optimization and convert this problem to per-time slot deterministic optimization, hence,enabling the time-independent decision making.
2. DADM follows a low-complexity design that schedules the requests and controls the data rate sequentially without loss of optimality, by observing the feedback from requests arrivals, queuing dynamic at providers, and links over the time slots.
3. The authors prove the [O(V), O(1/V)] optimality-queue length tradeoff can be achieved by DADM.

However, pure transmission solutions cannot be directly applied to CLS. Merely considering the path quality may flood a large amount of data into only a few transcoding nodes experiencing good network conditions and lead to the overload of those nodes or, even worse, to the collapse of the CLS system.