Towards Tactile Internet in Beyond 5G Era: Recent Advances, Current Issues and Future Directions

I. INTRODUCTION

• 5G targets include 20 Gbps peak data rates, 100 Mbps cell-edge rates, 10 Mbps/km^2 areal capacity, and 1 ms round-trip latency; TI also depends on ultra-low latency and five-nines reliability.These requirements motivate the paper’s focus on the research challenges of supporting TI over wireless systems.Source: §I · p. 1

A. Vision for TI and Recent Advances

• TI shifts communication from content delivery to real-time steering and control, transmitting touch and actuation alongside audiovisual information for human-machine and machine-to-machine interactions.Applications include remote surgery with touch and force feedback, road safety, teleoperation, AR/VR, remote driving, and eHealthcare.Source: §I · p. 1, A. Vision for TI and Recent Advances · pp. 1–2
• Existing infrastructure is insufficient for TI’s stringent latency, reliability, security, and availability needs, which also depend on human perception and reaction time.The paper points to distributed edge/cloud architectures and redesigned resource allocation, feedback, interference management, and MAC as potential responses.Source: A. Vision for TI and Recent Advances · p. 2
• TI also requires advances in tactile sensors and signal processing, wireless transmission, and standardization; ITU has published a Technology Watch report, while IEEE P1918.X and IIC efforts address related frameworks and low-latency industrial IoT.The IEEE family includes work on TI architecture, codecs, AI, and MAC.Source: A. Vision for TI and Recent Advances · p. 2

B. Review of Related Overview/Survey Articles

• Prior surveys cover IoT, mMTC/M2M, and URLLC technologies, applications, and challenges; other reviews address M2M scheduling, IoT analytics, and short-packet transmission.Table I classifies the related survey and overview works across IoT/mMTC and TI topics.Source: B. Review of Related Overview/Survey… · pp. 2–3, Table I · p. 3
• Latency-focused surveys review reduction technologies, delay-source-based technique classifications, and URLLC requirements and control-channel enhancements for LTE Releases 14 and 15 and 5G NR.Source: B. Review of Related Overview/Survey… · p. 3
• Existing TI overviews address application scenarios, reliability and latency in FiWi-enhanced LTE-Advanced, and 5G/TI requirements and architecture, while haptic and wireless AR/VR surveys cover their own techniques, challenges, and case studies.Haptic coverage includes quality metrics, compression, teleoperation, and communication protocols; the wireless VR work presents case studies on resource allocation and computing, proactive versus reactive computing, and AR-enabled self-driving vehicles.Source: B. Review of Related Overview/Survey… · p. 3

C. Contributions of the Paper

• The survey aims to address the lack of a holistic view of wireless TI and tactile-traffic challenges in complex wireless environments.Its objectives include reviewing recent advances and proposing enablers and future research directions.Source: C. Contributions of the Paper · p. 3

II. Tactile Internet in the Beyond 5G Era

• The proposed framework covers TI architecture, technical requirements, application areas, and the feasibility of existing technologies.The framework identifies ultra-responsive and ultra-reliable connectivity, edge intelligence, efficient tactile-data transmission, and low-complexity processing as key requirements.Source: C. Contributions of the Paper · p. 3, §II · p. 4

IV. Wireless Augmented and Virtual Reality

• The paper reviews haptic communications and wireless AR/VR, including their challenges, requirements, enablers, and use cases.The haptic-communications review includes a teleoperation example.Source: C. Contributions of the Paper · pp. 3–4, D. Paper Organization · p. 4

V. Autonomous, Intelligent and Cooperative Mobility Systems

• Autonomous, intelligent, and cooperative mobility systems are reviewed across automotive V2X, cooperative driving, drones, and mobile AR/VR.The review addresses challenges, enablers, and example use cases.Source: C. Contributions of the Paper · p. 4, D. Paper Organization · p. 4

VI. Enabling Technologies for Tactile Internet

• The paper identifies wireless TI challenges and discusses enabling technologies across physical/MAC and network layers, including cloud-level techniques.The proposed technologies also include network-edge approaches.Source: C. Contributions of the Paper · p. 4, D. Paper Organization · p. 4

VII. Security and Privacy for TI Applications

• Security and privacy threats to TI applications are reviewed alongside state-of-the-art work and promising enhancement techniques and architectures.The paper separately identifies open research issues and recommends future directions.Source: C. Contributions of the Paper · p. 4, D. Paper Organization · p. 4

D. Paper Organization

• The paper structure assigns Section VIII to open issues and future recommendations, followed by the conclusion in Section IX.Section III covers haptic communications, while Section IV addresses wireless augmented and virtual reality.Source: D. Paper Organization · p. 4

A. Tactile Internet Architecture

• TI adds transmission of touch and actuation information to conventional audiovisual communication.Source: A. Tactile Internet Architecture · p. 5

A. Review of Key Challenges

• The section identifies physical/MAC-layer solutions as one of the review’s key-challenge areas.The supplied passage provides the subsection heading only.Source: A. Review of Key Challenges · p. 5

B. Enabling Technologies

• The enabling-technologies section identifies network/cloud-level solutions as a solution area.The supplied passage provides the subsection heading only.Source: B. Enabling Technologies · p. 5

D. SDN-based Design

• The wireless TI system uses a two-way network link between a human-operated master domain and a controlled remote domain to exchange commands and haptic, audio, and visual feedback.The network domain may include an Internet/core network, a RAN, and a tactile support engine; the paper focuses on this communication network.Source: D. SDN-based Design · pp. 5–6, Figure 3 · p. 7

B. Key Technical Requirements

• TI applications commonly require about 1 ms end-to-end latency; a smart-factory example calls for 99.999% reliability at about 1 ms latency.Reliability may be improved with concurrent links or paths, higher SNR margins, and stronger channel coding; multiple paths depend on channel dynamics and CSI availability.Source: B. Key Technical Requirements · p. 7
• TI requirements also include edge AI/ML for activity prediction and caching, tactile encoding and audiovisual feedback, and security and privacy under strict latency constraints.The passage identifies low-overhead physical-layer security, secure coding, and reliable, low-latency legitimate-receiver identification as areas for investigation.Source: B. Key Technical Requirements · p. 7

C. Application Areas

• TI applications range from haptic communication, AR/VR, and remote surgery to exoskeletons, education, vehicles, industrial teleoperation, smart grids, and smart cities.IEEE 1918.1 use cases include immersive VR, teleoperation, automotive, drones, interpersonal communication, live haptic broadcast, and cooperative automotive driving; IVR haptic feedback latency is < 5 ms slave-to-master, while master-to-slave video and audio require < 10 ms and haptic feedback 1–50 ms.Source: C. Application Areas · pp. 7–8
• Wireless teleoperation enables interaction with real or virtual objects in distant, inaccessible, or hazardous environments using motion, vibration, texture, and force feedback.Unlike conventional short-distance wired systems, wireless teleoperation must meet strict latency and reliability requirements for timely remote interaction.Source: C. Application Areas · p. 8
• Tactile Internet can support immersive entertainment, gaming, live haptic broadcasting, and interactive education through real-time audio, video, and haptic transmission.These applications require a multimodal auditory, visual, and haptic human-to-machine interface with very low end-to-end latency.Source: C. Application Areas · p. 8

3. Autonomous, Intelligent and Cooperative Mobility

• Mobility TI spans autonomous and remote driving, platooning, coupled trains, and UAVs, all requiring stringent latency, reliability, and stability control.V2V and V2I communications are essential to future self-driving capabilities.Source: §3 · p. 8
• Other TI uses include wireless NCSs, smart energy, tactile robots, and industrial automation, where reliable low-latency control enables responsive or flexible operation.NCSs face instability from wireless delays and packet losses; TI may provide reliable ultra-low-latency connectivity. Smart grids need ultra-responsive control, while tactile robots enable remote physical interaction with haptic feedback and industrial systems must address reliability, latency, energy, and security requirements.Source: §3 · p. 8

D. Feasibility Analysis of Existing Wireless Technologies

• Factory studies show OFDM-based 5G can support sub-millisecond, highly reliable links, while LTE/5G results highlight the need for a new radio interface for stringent TI requirements,.Spatial diversity and short transmission intervals without retransmissions can yield very low error rates and delays under fast fading, interference, antenna correlation, and channel-estimation errors. Conventional wireless technologies designed for HTC are generally insufficient for TI’s latency and reliability needs.Source: D. Feasibility Analysis of Existing… · p. 9
• Current video systems’ 50–400 ms glass-to-glass delay makes them unsuitable for TI, motivating lower-delay video for teleoperation, connected driving, and wireless AR/VR.The reported delay varies with the separation between endpoints and the number of networks traversed.Source: D. Feasibility Analysis of Existing… · p. 9
• Conventional 2 × 2 and 4 × 4 MIMO configurations do not meet stringent reliability needs; macroscopic diversity and interference management may improve SINR outage, subject to diversity–multiplexing tradeoffs.Source: D. Feasibility Analysis of Existing… · p. 9
• An SDR broadband system demonstrated about 1 ms round-trip delay, but only on a single link without interference or queueing and protocol delays.Source: D. Feasibility Analysis of Existing… · p. 9
• Wi-Fi DSSS and OFDM respond differently to synchronization offsets, while HCCA analysis provides latency-based parameter selection; TI also needs load-aware MAC protocols [85],.The work identified operating limits for ultra-low-latency Wi-Fi and derived a closed-form expression for tactile-user-to-access-point queuing latency.Source: D. Feasibility Analysis of Existing… · p. 10
• For TI healthcare, a fixed-length SmartBAN transmission mechanism can provide less than 1 ms downlink delay, unlike conventional exhaustive transmission.The study modeled downlink delay and examined how transmission duration affects delay and energy performance.Source: D. Feasibility Analysis of Existing… · p. 10

III. HAPTIC COMMUNICATIONS OVER WIRELESS TI

• Haptic communication closes a bidirectional control loop between a user and remote environment, exchanging commands and haptic feedback alongside audio and video.Kinesthetic feedback conveys velocity, force, position, and torque, while tactile feedback conveys texture, surface, and friction.Source: §III · p. 10

A. Key Challenges

• Wireless haptic TI requires ultra-high reliability and about 1 ms end-to-end latency because delayed or lost critical feedback can destabilize control.Latency optimization spans protocol layers, air interface, backhaul, hardware, and core Internet; shorter TTIs require more bandwidth, and finite light speed limits endpoint separation. TCP reliability incurs high overhead and latency, while UDP reduces overhead at a reliability cost; a 3-DoF payload is 6 bytes versus a 40-byte IPv6 header.Source: A. Key Challenges · pp. 10–11
• Haptic communications need prioritized, symmetric uplink and downlink resources at a guaranteed minimum constant rate while coexisting with H2H and machine-type traffic.Flexible cross-layer allocation, adaptive flow management, and on-demand network slicing are proposed for this coexistence challenge.Source: A. Key Challenges · p. 11
• Integrating haptic, visual, and auditory feedback requires multiplexing across protocol layers to manage cross-modal asynchrony and heterogeneous latency, rate, and sampling needs.Existing suggested schemes are application-layer based; further work is needed for dynamically varying wireless environments.Source: A. Key Challenges · p. 11
• Multi-user haptic collaboration requires a peer-to-peer overlay, but overlay and IP routing can further increase end-to-end latency.Source: A. Key Challenges · p. 11
• Area-based touch sensing and actuation must extend beyond single-point devices to body surfaces, increasing data rates and changing how data loss is perceived.Source: A. Key Challenges · p. 11
• Wireless haptic TI needs novel QoE and QoT metrics and stable control despite time-varying delays and packet losses.QoE captures interaction transparency relative to local manipulation, whereas QoT measures task accuracy; mapping network QoS to haptic processing is needed for joint communication, compression, and control optimization.Source: A. Key Challenges · pp. 11–12
• TI needs standardized haptic codecs for 1 kHz sampled signals and practical lightweight displays that provide both cutaneous and kinaesthetic feedback.Sensor design must balance spatial resolution, scan time, sensitivity, and placement; larger arrays increase sensing time and power consumption.Source: A. Key Challenges · p. 12

• Joint uplink/downlink allocation must account for bidirectional coupling, bounded delay for control-loop stability, and a minimum rate throughout each haptic session.Source: B. Requirements and Potential Enablers · p. 12
• [6] formulated haptic power/RB allocation under uplink/downlink access constraints, decomposed it via optimal power control into binary integer programming, and used a low-complexity joint-scheduling heuristic.Source: B. Requirements and Potential Enablers · pp. 12–13
• [56] Hap-SliceR uses reinforcement learning to dynamically size slices against changing network dynamics and utility requirements, with post-decision state learning to improve efficiency.Source: B. Requirements and Potential Enablers · p. 13
• [57] surface classification with haptic adjectives was more accurate when it combined visual and physical interaction signals.Source: B. Requirements and Potential Enablers · p. 13
• [58] validated impedance control theoretically, in simulation, and experimentally; sending only force information performed similarly to standard bilateral control.Source: B. Requirements and Potential Enablers · p. 13
• [59] experiments showed Internet haptic-data transfer feasible, with limitations tied to geographic distance and network conditions; transport needs include reliability, prioritization, and congestion/flow control.Source: B. Requirements and Potential Enablers · p. 13
• Closed-loop surface haptics needs strict latency for immediate force or sensor updates, unlike open-loop systems without a global control loop.Source: B. Requirements and Potential Enablers · p. 13

C. Use Case Example: Teleoperation Systems

• Teleoperation systems combine a human-operated master, communication channel, and slave actuator, with control modes shaped by delay; they require about 1000 or more haptic packets per second and are sensitive to loss and latency.Even small delays or packet loss can destabilize bilateral teleoperation and degrade service quality.Source: C. Use Case Example: Teleoperation… · pp. 13–14
• Haptic data reduction and compression are proposed to limit teleoperation network-resource use while respecting strict stability and delay requirements.Reviewed methods include perceptual deadband, velocity-adaptive deadband, predictive coding, and event-based coding.Source: C. Use Case Example: Teleoperation… · p. 14

IV. WIRELESS AUGMENTED AND VIRTUAL REALITY

• AR overlays content on the real world, while VR creates an immersive experience; both need massive information flows and real-time interaction, including 3UH video requirements and careful tracking-delay and accuracy management [63, 62, 61].VR video requires ultra-high data rate, responsiveness, and reliability; motion-to-photon latency is normally 10–20 milliseconds, and tracking and video-generation delays also affect QoE.Source: §IV · pp. 14–15
• A proposed MIMO-OFDM 5G MAC scheduler supports multiple VR users by differentiating video frames, adapting links to dynamic BLER targets, and selecting users by aggregate delay-capacity utility.The design targets the maximum number of users while maintaining their 3UH QoE.Source: §IV · p. 14
• For cellular VR, a multi-attribute utility model jointly considers processing delay, tracking accuracy, and transmission delay in small-cell resource management.Small-cell base stations gather user tracking information uplink and send 3D images and audio downlink.Source: §IV · pp. 14–15
• An SDN architecture with multipath cooperative routing from multiple edge data centers is proposed to meet VR/AR low-latency and reliable-transmission needs.For 360° VR video, motion-to-photon latency should remain within 20 milliseconds to avoid user dizziness.Source: §IV · p. 15
• For interconnected VR in an ultra-dense 3D urban network, a two-stage cooperative transmission scheme mitigates inter-cell interference, with derived rate and latency-probability expressions validated numerically.The analysis uses stochastic geometry and densely deployed low-power nodes.Source: §IV · p. 15
• VR applications include immersive minimally invasive surgery and haptic rehabilitation, while cloud- or edge-server rendering is proposed to improve bulky HMD portability.The surgery system was validated with skilled surgeons; the rehabilitation system involved children with cerebral palsy and developmental dyspraxia.Source: §IV · p. 15

A. Key Technical Challenges

• Wireless VR research must address information-theoretic design, quality-rate-latency optimization, system scalability, and accurate user and object tracking.Topics include coding and compression under imperfect information, choosing content quality and payload size under network constraints, and improving eye tracking and object localization.Source: A. Key Technical Challenges · p. 15
• Wireless VR must also minimize energy use, balance headset and network computation, explore quantum computing, and preserve privacy automatically.Computation placement depends on bandwidth, latency, cost, and reliability trade-offs; privacy mechanisms should avoid burdening users or haptic devices.Source: A. Key Technical Challenges · p. 16

B. Potential Enablers

• Caching and collaborative edge-cloud processing can support wireless VR by avoiding repeated server access while combining cloud capacity with edge-side low-latency service.Cache placement can use user location, mobility, and social ties; cloud processing suits demanding workloads, while edge or fog processing supports location-aware, mobile, high-QoS applications.Source: B. Potential Enablers · p. 16
• Wireless VR enablers also include short-range device communication, contextual analytics, and behavioral and social data to improve network operation and user experience.M2M or D2D sharing among nearby users can ease congestion, while data-driven solutions may address switching among multiple screens.Source: B. Potential Enablers · p. 16

V. AUTONOMOUS, INTELLIGENT AND COOPERATIVE MOBILITY SYSTEMS

• The mobility discussion covers automotive V2X, cooperative automated driving, Internet of drones, and mobile AR/VR applications.These applications are considered within autonomous, intelligent, and cooperative mobility systems.Source: §V · p. 16

1) Automotive with V2X Communications:

• V2V alone may not meet latency needs at scale, so V2I can improve reliability and reaction time; virtualization examines spectrum allocation under network-switching constraints.The approach analyzes the trade-off between network switching rate and latency, while requiring isolation among virtual networks and accounting for added control signalling.Source: 1) Automotive with V2X Communications · p. 16
• Dense vehicular networks need low-latency, reliable V2X access, and a proposed NOMA-based hybrid centralized/decentralized scheme targets collision reduction and reliability.The passage identifies limited frequency use and congestion in LTE-based orthogonal multiple access networks as motivating issues.Source: 1) Automotive with V2X Communications · p. 16
• Collaborative driving and remote operation require high-volume, high-quality audiovisual and haptic data with very low latency, real-time sensing, and reliable feedback for safety-critical decisions.Existing V2V/V2I latency exceeds 10 ms, which the paper says is unacceptable for autonomous traffic; remote haptics may require tens to hundreds of degrees of freedom, and life-critical use includes remote driving support for trucks, shuttles, and road machines.Source: 1) Automotive with V2X Communications · pp. 16–17

2) Cooperative Automated Driving:

• Cooperative driving needs reliable, low-latency communication for sub-5 m safety gaps and roughly 10 kHz message rates, beyond current vehicular technologies under congestion.TI could support multi-vehicle perception and maneuvering by transferring sensor and driving-trajectory haptic data over V2V, V2I, or V2X links.Source: 2) Cooperative Automated Driving · p. 17
• Cooperative driving requires scalable resource allocation and dependable, coordinated, minimally correlated links; parallel redundant communications and consistency checks can help counter attacks.Platooning simulations found potential in combining synchronized communication slots with transmit-power adaptation.Source: 2) Cooperative Automated Driving · p. 17

3) Internet of Drones:

• TI could support drone delivery with ultra-reliable, low-latency links for dynamic route selection and collision avoidance, while transmitting haptic information alongside location, audio, and video.Managing traffic becomes important as delivery systems scale beyond the small numbers of drones used in trials.Source: 3) Internet of Drones · p. 17

4) Mobile AR/VR:

• Mobile VR can mitigate pain and psychological indicators such as heart rate, though it is less effective than costly wired HMDs, and may bring care to patients’ homes.The cited studies examine VR’s palliative effects and mobile-phone delivery of VR services.Source: 4) Mobile AR/VR · pp. 17–18
• Web AR offers lightweight, cross-platform mobile access, but energy, computation, and networking challenges remain; cloud offloading may add delay.Network slicing, D2D communications, and multi-access edge computing are identified as potential enablers.Source: 4) Mobile AR/VR · p. 18

A. Review of Main Challenges

• TI connectivity targets include latency below 1 ms, reliability above 99.99%, and application-dependent data rates up to Gbps–Tbps, alongside haptic codec, coordination, sensing, and multiplexing challenges.The review also calls for performance metrics suited to haptic systems and better balancing of XR quality, rate, latency, scalability, localization, and computation.Source: A. Review of Main Challenges · p. 18
• Short packets make metadata overhead and channel uncertainty harder to manage, while strict delay limits constrain retransmissions and congestion control; light speed also bounds tactile links to about 150 km.Suggested responses include short frames and codes, novel access and network architectures, energy-aware allocation, and edge ML that predicts actions; haptic reaction and multisensory feedback must also meet human timing limits.Source: A. Review of Main Challenges · pp. 18–19
• 31.6% more concurrent VR users were served than with fixed-BLER link adaptation and maximum-sum-capacity scheduling.The scheme uses video-frame differentiation, delay-based weighting, spatial-frequency user selection, and dynamic-BLER link adaptation.Source: A. Review of Main Challenges · pp. 19–20

1) Physical and MAC Layer Solutions:

• Joint cross-layer optimization of packet-error, queue-delay-violation, and packet-dropping probabilities minimized base-station transmit power; found all three probabilities were of the same order, while maximized conventional uplink sum-rate subject to tactile-delay constraints.The studies considered, respectively, reliability and delay in a TI system and an SCMA system carrying both tactile and conventional data.Source: 1) Physical and MAC Layer Solutions · p. 20
• Inter-cell coordination using narrowband muting, single-frequency networking, or macro-diversity with soft combining can improve transmission reliability without the longer delays associated with retransmissions.The techniques were evaluated for 5G URLLC in a typical indoor environment.Source: 1) Physical and MAC Layer Solutions · p. 20
• Optimizing uplink and downlink delay components alongside bandwidth reduced required total bandwidth by almost half versus no delay optimization, with a tradeoff between uplink and downlink bandwidth.The optimization targeted end-to-end delay and overall reliability under packet-loss and delay constraints.Source: 1) Physical and MAC Layer Solutions · p. 20
• Because haptic uplink and downlink sessions are coupled, QoS degradation in one direction affects the reverse direction; studied joint resource allocation for symmetric and perceptual coding.The cited work addressed resource allocation requirements for haptic communications in human-in-the-loop mobile networks.Source: 1) Physical and MAC Layer Solutions · pp. 20–21
• Traffic-state-aware bandwidth reservation saved bandwidth versus traffic-agnostic reservation, with savings increasing as traffic burstiness increased.The method classified packet arrivals into low- and high-traffic states using a Neyman-Pearson method.Source: 1) Physical and MAC Layer Solutions · p. 22
• For short-packet, low-latency communications, UFMC outperformed filtered CP-OFDM and FBMC in the reported comparison; FBMC incurred high time-domain overhead and degraded with short bursts.Finite-blocklength studies also found that uplink channel training becomes more beneficial as block length decreases and transmit antennas increase.Source: 1) Physical and MAC Layer Solutions · pp. 22–23
• End-to-end latency depends on interacting protocol layers and network segments, so single-layer optimization can harm other layers; analyzed and classified low-latency enabling technologies.For heterogeneous TI and healthcare traffic, time- and wavelength-division multiplexing in passive optical LANs with dynamic bandwidth and wavelength allocation was examined to meet end-to-end latency constraints.Source: 1) Physical and MAC Layer Solutions · pp. 21, 24
• To reduce latency across network segments, the paper identifies frame and waveform enhancements, edge caching and computing, advanced backhaul, SDN, NFV, machine learning, and MEC-enabled cloud services.These measures target transmission, backhaul, core-network, and transport latency, respectively.Source: 1) Physical and MAC Layer Solutions · p. 25
• LTE control overhead adds about 0.3 −0.4 ms per packet and retransmission takes around 8 ms, while eliminating retransmission time affects packet error rate.The paper identifies frame and packet structure, modulation and coding, transmission techniques, waveform design, and symbol detection as areas for latency reduction.Source: 1) Physical and MAC Layer Solutions · p. 25
• Soft uplink reservation outperformed conventional SR in latency, while burstiness-aware learning saved about 40%70% of bandwidth under required reliability and latency constraints,.The soft reservation reuses an uplink grant for subsequent transmissions without new device-to-BS signaling; the learning approach accounts for traffic-classification errors.Source: 1) Physical and MAC Layer Solutions · p. 25
• In C-RAN fronthaul, multi-link diversity with MAC-frame erasure coding supports reliable, low-latency transmission, and orthogonal resource allocation lowers average latency and error probability versus non-orthogonal allocation.Fronthaul latency requirements range from 55 µs to 1 ms depending on the CU–RU split.Source: 1) Physical and MAC Layer Solutions · p. 26
• For URLLC, 3GPP targets 10^-5 BLER with 1 ms radio latency; low-SNR control-channel reliability motivates improvements to channel estimation, CQI, SR, grants, and ACK/NACK detection.The models account for errors in both data and control channels.Source: 1) Physical and MAC Layer Solutions · p. 26
• Because feedback-based HARQ adds delay and signaling overhead while feedback-less one-shot transmission trades reliability against latency, an RL-based multi-armed bandit approach harmonizes the two.The approach uses the exploitation–exploration tradeoff to dynamically balance feedback-based and feedback-less transmissions.Source: 1) Physical and MAC Layer Solutions · p. 26

Software Defined Networking

• SDN-based softwarization was proposed to support mission-critical traffic through end-to-end reliability, multi-RAT support, and dynamic resource allocation across core and radio access networks,.The cited NFV-based SDN architecture runs 5G network functions as software components and uses SDN-like control of the data plane.Source: 2) Network Layer and Cloud-Level… · pp. 27–28

Network Virtualization

• For TI medical-imaging applications, virtualized sub-GHz radios combined with licensed mobile radio were proposed and tested using Ofcom TV whitespace.The requirements motivating this work include latency below about 1 ms and “five-nines” reliability; a flexible cloud-based RAN with dynamic slicing and traffic offloading was also proposed.Source: 2) Network Layer and Cloud-Level… · p. 28

Network Coding

• Network coding’s recoding capability enhances information reliability and balances the storage-repair traffic tradeoff without costly network-architecture investments; it can also improve TI communication latency and reduce packet-retransmission frequency.Source: 2) Network Layer and Cloud-Level… · pp. 27–28

Collaborative Edge/Fog and Cloud Computing/Processing

• MEC can reduce round-trip latency and offload computation, while a proposed micro-, mini-, and central-cloud hierarchy provides edge computing and requires optimizing cloud-unit counts,.Other proposals address fast migration of latency-sensitive applications between servers and task offloading among cloud, cloudlets, and neighboring robots, with frameworks for energy use, response time, and allocation delay.Source: 2) Network Layer and Cloud-Level… · pp. 28–29

Caching Techniques

• An energy-efficient hybrid caching scheme with a Zipf-based cache-replacement policy was proposed to reduce latency and improve energy efficiency for TI.Predictive caching was also identified as a way to address TI’s 100 km service-range limit when using fiber.Source: 2) Network Layer and Cloud-Level… · p. 29

Machine Learning (ML)/Artificial Intelligence (AI)

• AI can support tactile cognizance by predicting movement and action to compensate for latency, enable skill acquisition and replication, and predict reliable communication parameters.Understanding remote-terminal context and interactions helps deliver perceived real-time operation and haptic feedback quality and depth.Source: §6 · p. 29
• An artificial neural network was studied for learning uplink latency performance from diverse network features and making flexible bandwidth allocations to reduce latency.Source: §6 · p. 29
• The scalable intelligence-enabled platform combines information-centric networking for eMBB, URLLC, and mMTC scenarios with a separate plane for learning and decision-making.Its learning system supports generalization and personalization with conflicting, imbalanced, and partial data, and aims to remove traffic redundancy in 5G audio-visual TI.Source: §6 · p. 29

Security threats

• Jamming threatens 5G wireless TI: intelligent attacks target control or data channels, and control-signal jamming can disrupt packet transmission and reception.The passage identifies uplink and/or downlink control signals as targets whose jamming can cause data-channel packet failures.Source: §VII · p. 30
• IoT/TI security threats span the data-processing lifecycle, while privacy risks arise from device interaction, location services, low user awareness, and distributed data fusion,,.The passage groups security issues into terminal, transmission, processing, and management security, and stresses protecting data privacy across massive distributed IoT networks.Source: §VII · p. 30

1. Blockchain

• Blockchain can support privacy and access control through tamper-resistant records, while proposed cloud-based models trade reduced load or trusted exchange for added cost and resource use,,.The approach in combines encrypted cloud storage with ledger access policies that give data owners control; the multilayer model in reduces computational and network load but raises costs and resource use, while combines joint cloud computing and blockchain for trusted resource exchange and transaction records.Source: §VII · p. 31

2. Edge/Fog Computing

• Fog/edge TI requires trust verification for mobile, decentralized devices, secure node communications, privacy protection for sensitive data, and authentication across gateways,.Cloud reputation models may not transfer directly to dynamic fog environments, and nearby fog nodes can collect location, identity, and utility-usage information.Source: §VII · p. 31

3. Machine Learning; 4. SDN-based Design

• Machine learning can support access control, authentication, malware detection, and anti-jamming, but TI deployment must address constrained resources, low-latency reliability, and heterogeneous traffic,.The section calls for learning-based authentication, access control, and secure offloading or storage to counter physical- and MAC-layer attacks, jamming, eavesdropping, and smart attacks. SDN can simplify security-policy management using a global network view, detect malware-affected endpoints, and support unauthorized-device restriction and rapid detection of unauthorized transmissions,,.Source: §VII · pp. 31–32

VIII. OPEN ISSUES AND FUTURE RECOMMENDATIONS

• The section discusses open research issues and recommends future directions across multiple wireless TI topics.Source: §VIII · p. 32

A. URLLC in TI Applications

• Visual-haptic VR can use an eMBB slice for visual perception and a URLLC slice for haptic perception, which needs fixed rate and high reliability,.Immersive VR, teleoperation, and cooperative automated driving require haptic traffic alongside conventional audio and video.Source: A. URLLC in TI Applications · p. 32
• Energy-delay tradeoffs at 60 GHz, coexistence with eMBB and mMTC, multiplexing in both directions, and cross-layer TI traffic prioritization remain research issues.Continuous channel sensing in unlicensed spectrum limits maximum transmission time.Source: A. URLLC in TI Applications · p. 32
• Downlink preemptive scheduling lets URLLC overwrite eMBB resources, with puncturing indication to limit decoding impact; uplink pause-and-resume reallocates eMBB resources,.Intra-user multiplexing of eMBB and URLLC is also an issue when one user uses both services.Source: A. URLLC in TI Applications · p. 32
• Multiple links and paths can improve TI reliability by avoiding a single point of failure, while graph signal processing may address issues in large-scale wireless TI networks.The section identifies these techniques as promising future directions.Source: A. URLLC in TI Applications · pp. 32–33

B. Latency Reduction in Wireless TI Systems

• RAN latency reduction requires attention to mmWave impairments, short-packet transmission and channel design, asynchronous/non-orthogonal access, low-complexity antennas and detection, and latency-aware MAC.Future work includes addressing mmWave propagation and beamforming issues, investigating short-packet strategies and channel models, and designing access schemes robust to fading and interference.Source: B. Latency Reduction in Wireless TI… · p. 33
• Core-network latency research spans heterogeneous-resource orchestration, adaptive backhaul, caching and edge tradeoffs, cache protocols and handovers, and edge-side autonomous actions.The passage identifies computing, network, and storage orchestration, caching-related tradeoffs and protocol design including handover, and edge-side machine-learning actions that avoid waiting for network propagation.Source: B. Latency Reduction in Wireless TI… · p. 33

C. Dynamic Resource Allocation in TI Systems

• TI resource allocation remains underexplored; closed-loop applications need joint uplink/downlink design and coordinated communication, computing, and caching resources.Relevant resources include bandwidth, power, antennas, battery, processor and memory capacity, computation resources, and storage.Source: C. Dynamic Resource Allocation in TI… · p. 33
• Distributed edge/fog cooperation and collaborative edge-cloud processing are promising, while edge-node efficiency, task offloading, response time, and latency-aware user association require attention.The passage proposes cooperation among edge nodes with different energy and computing resources, and identifies offloading and response-time challenges alongside collaborative edge-cloud processing.Source: C. Dynamic Resource Allocation in TI… · p. 33
• SDN and network virtualization can support isolated, customizable slices for priority-based haptic scheduling; future work includes slicing, isolation, allocation, and scheduling techniques.These approaches are proposed to flexibly manage network and resource slices across devices and heterogeneous applications.Source: C. Dynamic Resource Allocation in TI… · p. 33

D. Haptic communications

• Haptic communications require symmetric joint uplink/downlink allocation, bounded delay, and minimum session rates, alongside reliability, responsiveness, flexible prioritization, sensing, and multiuser support.Other challenges include device and codec design, control-loop stability, cross-modal asynchrony among visual, haptic, and auditory modalities, and defining new performance metrics.Source: D. Haptic communications · pp. 33–34
• Time-varying delays and packet losses can destabilize practical haptic control; machine learning is proposed to improve control-loop stability.The passage also points to physical/MAC-layer and cloud/network-level solutions for these issues.Source: D. Haptic communications · p. 34

E. Wireless Augmented/Virtual Reality

• Wireless AR/VR must address architecture and interfaces, QoS, tracking, high-rate low-latency links, mobile compute and energy limits, and compatibility across diverse platforms.The passages also identify rate adaptation, network-layer reliability, motion prediction, transport adaptation, and SDN-based QoS as research issues.Source: E. Wireless Augmented/Virtual Reality · p. 34

F. Hardware, Computational and Network Architectural Issues

• Because a dedicated haptic network is economically infeasible, flexible B5G networks could use slicing with SDN, NFV, and self-organizing RAN solutions to support diverse services and coordinate across networks.Existing TI work often assumes both ends share one wireless network, while real deployments may span different networks.Source: F. Hardware, Computational and Network… · p. 34
• Tactile sensing requires suitable sensor sensitivity, parameter coverage, arrangement, spatial resolution, conformability, integration, and low-complexity processing of raw signals.The text calls for advances in nanotechnology and fabrication to integrate sensing units with signal-processing modules.Source: F. Hardware, Computational and Network… · p. 34
• TI computation must enable near-real-time haptic feedback, in-field processing to reduce ingress transmissions, and intensive AI processing and training at the tactile edge.An adaptive, energy-efficient computing-box concept with ultra-short-range optical links between chips is proposed as a way to increase computing capacity per unit area.Source: F. Hardware, Computational and Network… · p. 34

G. Joint Design of Control and Communication Systems

• Teleoperation requires joint network-aware control and control-aware networking to balance QoE with stable global control despite variable delay, including dynamic control switching.The communications and control communities have handled these issues independently, and the text calls for greater interaction between them.Source: G. Joint Design of Control and… · p. 35
• Cooperative driving needs joint design of advanced control and efficient inter-vehicle communication protocols for safe, stable platooning and reliable, low-latency information delivery.Platooning control also depends on real-time information from neighboring vehicles and maintaining inter-vehicle spacing.Source: G. Joint Design of Control and… · p. 35

H. Advanced Signal Processing and ML for TI

• TI signal processing and ML must extract useful information from noisy, high-dimensional tactile data and address dynamic channels, short stationarity, distributed data, and computation and energy limits.Edge intelligence, caching, predictive ML modules, and hybrid data-driven/model-driven methods are proposed to reduce congestion and latency and improve haptic stability and communication range [54, 113].Source: H. Advanced Signal Processing and ML… · p. 35

IX. CONCLUSIONS

• Wireless TI requires ultra-low latency, ultra-high reliability, high data rates, energy and spectral efficiency, and network throughput across diverse applications; the survey reviews its paradigms, enablers, security, and future research.The survey covers haptic communications, wireless AR/VR, and autonomous, intelligent, and cooperative mobility systems.Source: §IX · p. 35