With the rapid development of smart grids, Demand Response Management (DRM) is expected to become the most effective and reliable solution to reduce peak load in electricity grids for adaptation of fluctuating electricity supply. It aims to shape the users' electricity loads by exchanging information in real time between the power utility and demand sides. Thus, a real-time communication mechanism needs to be designed for information transmission in demand response. In this paper, we propose a task-oriented communication mechanism to balance the electricity load and supply as quickly as possible in real-time demand response. The cumulative mean square error (MSE) between the electricity load and supply is defined as the performance metric of the designed communication mechanism. On this basis, both centralized direct control and distributed control through electricity price have been considered in demand response. Under these control mechanisms, task-oriented incremental source coding is introduced for the minimization of cumulative MSE between the electricity load and supply. Finally, numerical results are presented to show the potential of our proposed incremental source coding mechanism for real-time demand response.

Extremely low-latency communication has attracted considerable recent attention because it holds the promise of supporting emerging real-time applications such as autonomous driving, smart grids, and Industrial Internet of Things (IIoT). Owing to the limited bandwidth in wireless environments, the sub-packets or even bits have to be transmitted successively, thereby inducing non-negligible delay-induced cost for real-time remote monitoring, estimation, decision making, and control. In this paper, we present a unified incremental decoding framework for real-time applications, the costs of which are extremely sensitive to the latency of each individual sub-packet or bit. In contrast to conventional methods, in which a decision is made after fully decoding the entire packet, the incremental decoding strategy allows monitors or actors to make their decisions in real time based on partially received packet. By this means, there is no need to wait for the whole packet to be decoded, thereby reducing the delay-induced costs substantially. To minimize cumulative cost during the real-time monitoring and control, we design source coding and decision making algorithms jointly, in which a backward induction property is found. Furthermore, we conceive a dynamic programming algorithm for a given source codebook to significantly reduce the cumulative decision costs while maintaining low computational complexity.

Extremely low latency communications has attracted considerable attention because of its potential in emerging real-time applications such as autonomous driving, smart grids, and Industrial Internet of Things (IIoT). Incremental decoding, which allows a remote controller to make real-time decisions without waiting for the whole packet or codeword to be decoded, may significantly reduce the latency-induced cost, especially when the bandwidth is very limited. In this paper, we are interested in the incremental decoding for the real-time control of a Markovian system. In contrast to the conventional Markov decision process, in which the decision maker receives the state information immediately, we consider a remote control scenario in which the transmission delay of its remote state information is non-negligible. To maximize the average reward in such Markovian systems, we conceive a joint incremental decoding and Markov decision making policy. In particular, the instantaneous codebook is determined by the controller's local state, which is also known by the sensor, while the controller can make its decision in real time based on partially received packet. Furthermore, we demonstrate the policy's potential by applying it into automated highway driving scenario, the performance gain of which has been revealed by numerical results.

In the emerging Industrial Internet of Things (IIoT), real-time control is expected to play a key role. How to minimize the cost to be paid due to the transmission delay of digital signaling in real-time control system becomes a challenging problem. In this paper, we study incremental decoding based low latency communication for a real-time control system. The real-time control action will be updated, whenever a new bit is received instead of the entire codeword. In other words, when the controller obtains partial information of the digital signaling, it executes the control action immediately instead of waiting until the complete information is obtained, which is in contrast to the conventional real-time control. Our aim is to minimize the expected cumulative control cost over the control process by joint design of source coding and its corresponding real-time control scheme. To this end, we first show a recursive structure that reveals the relationship of minimal expected cumulative control cost among two adjacent decision epochs. Based on such structure, the optimal solution can be obtained by a recursive algorithm presented by us. Finally, our numerical results also demonstrate that the cumulative control cost over the control process can be significantly reduced by implementing the source codebook we proposed, compared with traditional source coding.

Energy harvesting (EH) aided communications hold a great potential in the design of green communication systems for their high energy efficiency. However, the random power supply due to EH may cause an intolerable delay in data transmission. To overcome this, a Reliable Energy Source (RES) is desired to provide transmission power when the large delay is induced. In this paper, we study the delay-optimal scheduling policy for EH aided communications with the constraint of average power provided by RES. More specifically, the delay-minimal scheduling is obtained through the two-dimensional Markov chain modeling and linear programming (LP) formulation. To further reduce the computational complexity, we present a value iteration algorithm, based on which we not only reveal a threshold-based structure of the delay-optimal scheduling policy for EH-aided communications with large-capacity batteries, but also conceive a low complexity policy that is asymptotically optimal. For EH-aided communications with finite-capacity batteries, we present a unified framework based on large deviation theory. The non-asymptotic framework demonstrates that the delay-power tradeoff curve of the low complexity scheduling policy is capable of converging to that of the delay-optimal policy exponentially as the capacity of the battery increases.

Real-time monitoring and remote control of stochastic systems have attracted considerable attention due to their potential in task-oriented communications and industrial Internet of Things (IIoT). How to achieve ultra high-freshness in real-time monitoring and remote control becomes a challenging problem. In this paper, we are interested in the freshness oriented source coding with incremental decoding. This is contrast to con-ventional source encoding/decoding, in which a random sample is estimated after its entire codeword is received. Incremental decoding, however, allows the real-time estimation of a random sample once a new bit or channel coding block is decoded in the physical layer. Its source codebook is then optimized, based on which we further conceive a real-time decision policy. Our policies minimize the average mean square error (MSE) or decision cost by judiciously designed codebook for source encoding. Numerical results show that the incremental decoding substantially reduces the MSE and decision cost in real-time monitoring.

Low-latency and energy-efficient wireless communication holds the potential of enabling the industrial internet of things (IIoT), automatic driving, and telesurgery. Cross-layer scheduling, which is aware of both the channel and queue states, has attracted considerable attention recently because it is capable of reducing the average delay substantially while meeting a given average power constraint. As a result, there has been considerable work, in which joint channel and buffer aware scheduling is formulated as a constrained Markov decision process (CMDP). In general, the optimal solution to a CMDP problem is characterized by the stationary probability of actions, yielding probabilistic cross-layer scheduling with possibly high complexity in practice. In this paper, we are interested in the low-latency and energy-efficient stationary cross-layer scheduling policy for wireless channels with continuous channel states, e.g. Rayleigh fading. It is interestingly shown that a deterministic cross-layer scheduling policy can achieve the optimal tradeoff between the average delay and power. In other words, the signaling complexity of cross-layer scheduling can be significantly reduced without causing sub-optimality. Simulation results also demonstrate that our work provides a low-complexity solution for low-latency and energy-efficient wireless communications.

Energy harvesting (EH) aided millimeter Wave (mmWave) communications hold a great potential in the design of next generation wireless networks for their high data rate and energy efficiency. However, both the random power supply due to EH and the random blocking nature of mmWave channels induce severe queueing delay, thereby damaging the quality of service (QoS). To overcome this problem, in this paper, a cross-layer probabilistic scheduling is proposed for EH aided mmWave communications with random blocking channel. Our aim is to minimize the average delay while assuring that renewable energy is fully exploited. To achieve this goal, we formulate a two-dimensional Markov chain consisting of a data packet queue and a virtual queue of harvested energy. Based on the two-dimensional Markov chain, the average delay and the renewable power utilization of probabilistic scheduling policy are derived. On this basis, we obtain the delay optimal scheduling policy through a linear programming (LP) problem. More importantly, the structure of the delay optimal scheduling policy is shown to be threshold-based. Simulation results also validate that the threshold-based scheduling is capable of attaining significant QoS and energy efficiency gain.