Cognitive Hierarchy Theory for Distributed Resource Allocation in the Internet of Things

Abstract

In this paper, the problem of distributed resource allocation is studied for an Internet of Things (IoT) system, composed of a heterogeneous group of nodes compromising both machine-type devices (MTDs) and human-type devices (HTDs). The problem is formulated as a noncooperative game between the heterogeneous IoT devices that seek to find the optimal time allocation so as to meet their qualityof-service (QoS) requirements in terms of energy, rate and latency. Since the strategy space of each device is dependent on the actions of the other devices, the generalized Nash equilibrium (GNE) solution is first characterized, and the conditions for uniqueness of the GNE are derived. Then, to explicitly capture the heterogeneity of the devices, in terms of resource constraints and QoS needs, a novel and more realistic game-theoretic approach, based on the behavioral framework of cognitive hierarchy (CH) theory, is proposed. This approach is then shown to enable the IoT devices to reach a CH equilibrium (CHE) concept that takes into account the various levels of rationality corresponding to the heterogeneous computational capabilities and the information accessible for each one of the MTDs and HTDs. Simulation results show that the CHE solution maintains stable performance. In particular, the proposed CHE solution keeps the percentage of devices with satisfied QoS constraints above 96%for IoT networks containing up to 10,000 devices without considerably degrading the overall system performance in terms of the total utility. Simulation results also show that the proposed CHE solution brings a two-fold increase in the total rate of HTDs and deceases the total energy consumed by MTDs by 78% compared to the equal time policy. N. Abuzainab and W. Saad are with the department For example, the performance of MTDs that require ultra low latency or HTDs that require high data rates can can be severely affected by collisions. Thus, there is a need to design a new, distributed IoT multiple access scheme that can satisfy the requirements of devices with strict QoS constraints. A. Related WorksThere has been significant recent interest in developing resource allocation mechanisms suitable for the IoT such as in [8]-[14], [16], [17]. Centralized scheduling schemes for IoT LTE networks are proposed in [8]-[11]. In [8], a resource management scheme that dynamically allocates time resources between MTDs and HTDs based on current traffic conditions and QoS requirements. The works in [9] and [10] propose schemes that allocate the LTE resources to MTDs and HTDs based on a bipartite graph. In [11], the authors propose two seperate uplink scheduling schemes for HTDs and MTDs in an LTE system based on channel conditions and delay requirements while taking fairness into account. Other works such as in [12] and [13]adopted game-theoretic approaches for distributed resource allocation problems in the IoT. The authors in [12] study the problem of throughput maximization of MTDs under random access. However, in [12], devices are considered of equal ca...

Minimum Age of Information in the Internet of Things With Non-Uniform Status Packet Sizes

“…Based on J(X, w), we further define: 4 The upper limits of the AoI at the device d,k and the AoI at the receiver r,k guarantee the system state space to be finite, based on which these results in [29] can be used to prove Lemma 1. 5 The notation…”

“…Here, Pr[X ′ k |X k , w k ] is given by (5), θ k andV k (X k ) are the per-device average the AoI at the receiver and the per-device value function under policyπ, respectively.…”

“…Ensuring a seamless operation of real-time Internet of Things (IoT) applications [2]- [5] requires a timely delivery of status information collected from a variety of sensors that monitor physical processes. To characterize this timeliness of information update, the notion of age of information (AoI) has been recently proposed [6].…”

Multimedia Crowdsourcing With Bounded Rationality: A Cognitive Hierarchy Perspective

“…4(b), we plot the optimal reward R CHE * 1 and the optimal quality requirements Q CHE * against N H , respectively. 27 In Fig. 4(c), we plot the optimal quality requirements Q N E * against N H to compare the FR and BR models.…”

“…Fig. 4(a) shows that the 27 The analysis of R CHE * 2 and R CHE * reward for task 1 drops a few times in this process, corresponding to the instances where the competitions among tasks increase and the number of workers potentially interested in task 1 reduces.…”