Source-linked AI summary
Realizing the Tactile Internet: Haptic Communications over Next Generation 5G Cellular Networks
Adnan Aijaz, Mischa Dohler, A. Hamid Aghvami, Vasilis Friderikos, Magnus Frodigh
TL;DR
The paper addresses how to realize a Tactile Internet capable of real-time haptic communication under stringent latency, reliability, and stability requirements. It reviews haptic and networking challenges for 5G and proposes architectural, networking, edge-intelligence, and codec directions. It identifies 1-ms round-trip latency, enhanced haptic perception, and standardized haptic codecs as central requirements and open challenges.
Problem
Realizing the Tactile Internet requires haptic communications and 5G cellular networks to support real-time touch and actuation with stringent latency, reliability, and stability requirements.
Method
The article reviews haptic and networking requirements, examines 5G design challenges, and proposes networking, edge-intelligence, and codec directions.
Results
The paper identifies 1-ms round-trip latency, enhanced haptic perception through artificial intelligence and predictive analytics, and standardized haptic codecs as central design requirements and research challenges.
Takeaways & Limitations
Realizing the Tactile Internet requires coordinated changes to wireless-edge air interfaces, network architectures, haptic perception methods, and codec standardization.
Abstract
from arXiv · showhide
Prior Internet designs encompassed the fixed, mobile and lately the things Internet. In a natural evolution to these, the notion of the Tactile Internet is emerging which allows one to transmit touch and actuation in real-time. With voice and data communications driving the designs of the current Internets, the Tactile Internet will enable haptic communications, which in turn will be a paradigm shift in how skills and labor are digitally delivered globally. Design efforts for both the Tactile Internet and the underlying haptic communications are in its infancy. The aim of this article is thus to review some of the most stringent design challenges, as well as proposing first avenues for specific solutions to enable the Tactile Internet revolution.
I. INTRODUCTION
The Tactile Internet extends Internet connectivity toward real-time transmission of touch and actuation, enabling remote haptic control and new skill- and labor-delivery applications. Realizing it requires ultra-low latency, high reliability, and coordinated haptic and networking designs.
- Motivation: Applications include remote monitoring and surgery, controlled exoskeletons, remote education, remote driving, and industrial servicing.
- Requirements: Because it serves critical societal functions, the Tactile Internet requires ultra-reliable connectivity, high capacity, and very low end-to-end latency.Insufficiently low latency can produce cyber-sickness for tactile users.
- Requirements: The 1ms-Challenge targets a 1 ms round-trip latency with about 1 ms of outage per day, compared with roughly 20 ms latency in 4G networks.This target is identified as a key requirement for 5G mobile networks supporting real-time haptic interaction.
- Motivation: The Tactile Internet will transport touch and actuation in real time, with haptic communications as its primary application alongside audio, video, and data.Haptic information includes kinesthetic feedback such as force, torque, position, and velocity, as well as tactile feedback.
- Scope and approach: The article reviews haptic and networking requirements, translates them into 5G cellular design challenges, and offers practical recommendations for addressing them.
II. TOWARDS A TACTILE INTERNET ARCHITECTURE
The Tactile Internet architecture links a human-operated master domain to a remote slave domain through a network domain that carries bilateral haptic communication. Edge-located tactile support engines add artificial-intelligence capabilities for system stability.
- Architecture: The end-to-end architecture comprises master, network, and slave domains.
- Master domain: The master domain contains a human operator and haptic human-system interface that converts human input into haptic input and controls the slave domain.
- Network domain: The network domain provides bilateral communication that kinesthetically couples the operator to the remote environment.
- Slave domain: The slave domain contains a teleoperator that interacts with remote objects and exchanges energy with the master through command and feedback signals.This exchange closes a global control loop.
- Edge support: Tactile support engines near the network edge provide artificial-intelligence capabilities intended to stabilize the overall system.
- Service-medium relationship: Haptic communications are the primary service running over the Tactile Internet, which serves as their communications medium.The paper compares this relationship with VoIP and the Internet.
III. RESEARCH CHALLENGES FOR THE TACTILE INTERNET
Realizing the Tactile Internet requires advances in haptic devices, standardized codecs, multimodal integration, and the functional architecture supporting haptic transport.
- A. Haptic Devices: Existing haptic devices support touching, feeling, and manipulating real or virtual objects, but require higher degrees of freedom, network interfaces, lower cost, and combined kinesthetic-tactile feedback.Commercial devices with up to 6 degrees of freedom are cited as examples.
- B. Haptic Codecs: Haptic signals are typically sampled at 1 kHz, creating 1000 packets per second and motivating compression techniques for bandwidth-limited networks.
- B. Haptic Codecs: A standard haptic codec family should encode both kinesthetic and tactile information and support layered multimodal data for challenging wireless environments.The paper presents standardization as a fundamental challenge for scalability and universal uptake.
- Architecture: The architecture shown in Fig. 1 provides the medium for haptic transport.
C. Multi-Modal Sensory Information
Haptic communications must coordinate multiple sensory modalities, maintain stability and reliability in a closed control loop, and minimize protocol overhead. These requirements motivate joint communication-control design and protocol mechanisms tailored to haptic transport.
- C. Multi-Modal Sensory Information: Audio and visual feedback must accompany haptic feedback because the human brain integrates multiple sensory modalities.
- C. Multi-Modal Sensory Information: Cross-modal asynchrony arises because visual, auditory, and haptic modalities differ in sampling, transmission-rate, and latency requirements.A multiplexing scheme must exploit modality priorities and temporal integration.
- D. Stability for Haptic Control: Communication artifacts can destabilize the global control loop linking the human, cellular network, and remote environment, especially under time-varying delays and packet losses.
- E. Ultra-Reliability: Ultra-reliable connectivity is required because packet losses can produce undesirable haptic artifacts such as strong forces and erroneous surface roughness.Seven nines corresponds to an outage probability of 10^-7 and milliseconds of outage per day.
- E. Ultra-Reliability: Reliable haptic transport requires mechanisms beyond conventional ARQ and H-ARQ because retransmission delay conflicts with stringent timing requirements.
- E. Ultra-Reliability: Haptic session information has many parameters and must be exchanged reliably, although it is not subject to the hard delay constraints of the transport stream.
- E. Ultra-Reliability: For a 3-DoF stream, a packet carries only 6 bytes of payload, making IPv6 headers of 40 bytes a substantial overhead.
- E. Ultra-Reliability: Existing header compression can reduce 40-byte headers to less than 2 bytes, but may suffer on unreliable wireless links and introduce delay.
F. Ultra-Responsive Connectivity
The Tactile Internet’s 1 ms round-trip target requires coordinated optimization across air interface, protocol stack, hardware, transport, core Internet, and network architecture, while propagation speed imposes a separation limit.
- Physical-Layer Constraints: 33 µs packet duration is required for a 100 µs one-way Physical-layer transmission, but LTE’s approximately 70 µs OFDM symbols cannot meet this constraint.Shorter TTIs can reduce over-the-air latency but require higher available bandwidth.
- Latency Requirements: 1 ms round-trip latency requires optimizing every contributing segment, including air interface, protocol stack, hardware, backhaul, core Internet, and architecture.Air-interface delay is shaped by control- and user-plane latency; core Internet delay varies with queueing and routing policies.
- Physical-Layer Constraints: Finite speed of light sets the ultimate upper bound on the maximum separation between tactile ends.Hardware, protocol, and architecture advances can reduce delays but cannot remove this propagation limit.
G. Radio Resource Allocation
Haptic communications make radio resource allocation more demanding because bidirectional, low-latency traffic must coexist with other applications and collaborative users over shared cellular resources.
- Resource Coexistence: Radio resource allocation directly affects throughput, latency, reliability, QoS, and higher-layer performance when haptic and non-haptic traffic share cellular resources.The challenge arises from sharing resources among haptic applications, H2H traffic, and machine communications.
- Haptic Service Requirements: Haptic traffic requires priority resources, joint UL/DL allocation, and symmetric minimum constant rates because it supports bidirectional control.These requirements aim to maintain high tracking performance between master and slave domains.
- Resource Coexistence: Flexible, on-demand radio-resource management is needed for haptic and other vertical applications to coexist in 5G networks.The paper identifies novel allocation approaches as necessary to meet haptic requirements alongside diverse services.
- Collaborative Haptics: Collaborative multi-user haptic communications require a peer-to-peer overlay for coordinating users and maintaining overlay operations.Overlay routing may diverge from IP-level routing, introducing additional delays that challenge low-latency communication.
I. Area-Based Sensing and Actuation
Greater haptic immersion requires distributed or area-based sensing and actuation rather than single-point contact, while objective QoE evaluation remains underdeveloped alongside costly subjective testing.
- Area-Based Haptics: Single-point haptic end effectors do not match human perception of touch across surfaces, motivating distributed or area-based sensing and actuation.The paper cites the palm and other body regions as examples of surface-based touch perception.
- Objective Quality Metrics: Subjective haptic QoE testing is expensive and its credibility depends on tester selection, sample size, and environmental conditions.These factors make consistent subjective evaluation difficult.
- Objective Quality Metrics: Objective QoE evaluation is widely unaddressed, and accurately capturing haptic experience requires further development of mappings from service-delivery parameters.The cited passage identifies objective measurement as a needed direction rather than a completed solution.
A. Enabling Architecture for Network Slicing
The paper proposes a logical 5G architecture that uses a common programmable infrastructure, NFV, SDN, and SON to support flexible slicing for diverse vertical applications, while also outlining latency-oriented protocol and transport measures.
- Network Slicing: Separate haptic networks are considered infeasible because of CAPEX and OPEX, favoring one flexible 5G infrastructure shared across vertical applications.The proposed sharing model includes haptic, smart-grid, M2M, and V2V applications.
- Network Slicing: Network slices provide customizable connectivity services whose software-defined functions govern properties such as coverage, availability, and robustness.Slicing abstracts the common physical infrastructure for different application sectors.
- Enabling Technologies: NFV separates network functions from hardware, while SDN decouples control and data planes and enables software-based network-control programmability.Together they increase network abstraction and flexibility.
- Enabling Technologies: The proposed NFV-, SDN-, and SON-based architecture enables flexible, dynamic slicing of end-to-end network and service resources.VNFs run on common programmable infrastructure, with SDN and SON controllers programming core and radio-access networks.
- Latency Reduction: Tunable OFDM adapts parameters such as subcarrier spacing, FFT size, and cyclic prefix to reduce Physical-layer latency under suitable channel conditions.Small-delay-spread channels can support increased subcarrier spacing and reduced FFT and cyclic-prefix sizes.
- Latency Reduction: Air-interface latency can be reduced through fast RRC re-establishment, optimized random access, and alternatives to HARQ retransmissions for haptic traffic.HARQ’s retransmission delay is considered unsuitable for haptic communications, requiring other means of link-level reliability.
- Latency Reduction: Optical transport or full-duplex wireless backhaul are identified as ways to reduce backhaul delay.The wireless alternative is especially relevant in higher spectrum bands.
C. Achieving Ultra-Reliable Connectivity
Ultra-reliable haptic connectivity requires avoiding retransmission delays while exploiting multiple uncorrelated channels across frequency or space. The paper identifies inter-band aggregation and coordinated multi-point as ways to create such diversity.
- Frequency diversity over multiple uncorrelated links can approach carrier-grade reliability for haptic data and session-parameter exchange.The paper identifies inter-band spectrum aggregation and coordinated multi-point as mechanisms for creating multiple links.
- Because haptic latency constraints make time-domain diversity infeasible, HARQ cannot be applied in its conventional retransmission-based form.
- Frequency- or spatial-domain diversity can provide HARQ-like link-level reliability without retransmissions by exploiting multiple uncorrelated channels.
D. Virtualization-based Radio Resource Allocation Scheme
The proposed scheme virtualizes radio resources to support flexible slicing, isolation, and application-specific customization for haptic and other services. Its main unresolved issue is determining the optimal split between these applications.
- Virtualized radio resources enable flexible slicing, isolation, and customization across applications and devices, prioritizing tactile traffic and tactile-specific scheduling.
- Slicing: The scheme combines bandwidth-based and resource-based provisioning when initially allocating radio resources to different slices.
- Isolation: End-to-end isolation prevents channel conditions or user mobility in one slice from reducing resources across other slices.
- Customization: Resources are customized by service requirements, including persistent scheduling for haptic traffic when dynamic scheduling creates disproportionate control signaling.
- Slicing Period: Slice sizes are recalculated at each slicing period, with more frequent adjustment required in dynamic environments.
- Optimal slicing of radio resources between haptic and other applications remains future work.
E. Edge-Intelligence for Stability of Haptic Control
Latency and packet loss can destabilize haptic control, while edge intelligence is proposed to reduce latency and predictive analytics to mitigate its impact. The conclusion identifies wireless-edge redesign and haptic perception enhancement as central routes toward the 1-ms challenge.
- Latency and packet loss are primary wireless causes of instability in haptic control, with latency reported as more detrimental than packet loss.
- Edge caching and user-oriented traffic management reduce core-network congestion and end-to-end latency, supporting the Tactile Internet.
- The paper concludes that achieving 1-ms round-trip latency requires cutting-edge networking design alongside enhanced haptic perception.
- Artificial intelligence and predictive analytics are proposed to understand haptic actuation and enhance haptic perception.
- A standard haptic codec family is identified as a major research challenge for enabling advanced haptic features.