Dynamic Distributed TDMA Resource Allocation Algorithm for Robotic MANETs

Ⅰ. Introduction

Robotic Mobile Ad-hoc Networks (MANETs) have gained significant attention in recent years due to their potential in various mission-critical applications such as search and rescue, environmental monitoring, and military operations^[1] (Figure 1). These networks con- sist of autonomous robotic nodes that communicate with each other without relying on fixed infrastructure. The high mobility of robots leads to frequent changes in network topology, including dy- namic merging and splitting of network partitions^[2,3]. This dynamic nature poses substantial challenges for efficient resource allocation and reliable communi- cation.

Fig. 1.

Robotic Swarm MANET Scenario.

However, the inherent characteristics of MANETs, particularly in robotic networks, pose numerous chal- lenges for communication protocols. First, the high mobility of robots leads to frequent changes in net- work topology, which complicates the task of main- taining stable communication links. Additionally, the limited bandwidth and energy constraints of robotic systems make it imperative to optimize the use of communication resources. The unpredictable nature of tasks in robotic networks further adds to the complex- ity, as communication requirements can vary sig- nificantly over time, requiring flexible and dynamic resource allocation mechanisms.

Time Division Multiple Access (TDMA) is a popu- lar medium access control (MAC) protocol in MANETs due to its deterministic nature, where com- munication is scheduled in specific time slots, re- ducing the chances of packet collisions. However, tra- ditional TDMA schemes suffer from inefficiency in highly dynamic environments like robotic MANETs^[4,5]. Static slot allocation in conventional TDMA can lead to underutilization of bandwidth, es- pecially when nodes with low communication needs are assigned the same number of slots as more active nodes. Moreover, when network topology changes due to node mobility, the communication schedule must be frequently recalculated, leading to significant de- lays and overhead.

To address these issues, this paper proposes a D2TRA-RM: Dynamic and Distributed TDMA Resource Allocation for Robotic MANETs. Our ap- proach introduces a flexible slot allocation mechanism that dynamically adjusts based on the communication demands and network topology changes. Each node in the network is capable of independently requesting additional slots or releasing unused ones, thus ensur- ing that bandwidth is efficiently allocated to nodes with higher communication demands. This dynamic allocation allows the network to adapt in real time to changes in both traffic and topology, improving overall network performance in terms of throughput, latency, and bandwidth utilization.

The proposed algorithm incorporates several key innovations that distinguish it from traditional TDMA schemes. First, dynamic slot allocation enables nodes to request or release slots based on real-time commu- nication needs, effectively reducing bandwidth waste and ensuring that high-priority tasks receive sufficient resources. Second, the algorithm operates in a decen- tralized manner, where each node makes slot alloca- tion decisions independently, relying on local in- formation without the need for global synchronization, thus reducing overhead. In addition, priority han- dling for critical data ensures that time-sensitive in- formation is prioritized by preempting less critical communications, minimizing delays for high-priority messages. Finally, the algorithm is designed to effi- ciently reallocate slots in dynamic topologies, al- lowing nodes to adjust their slot assignments in re- sponse to frequent topology changes, thereby prevent- ing communication interruptions caused by node mobility.

In this paper, we provide a comprehensive evalua- tion of the proposed D2TRA-RM algorithm through extensive simulations using the NS-3 network simulator. We compare the performance of our dy- namic TDMA scheme with traditional static TDMA and a naïve Dynamic-Distributed TDMA scheme in various robotic network scenarios. Our results demon- strate that the proposed algorithm achieves a 198.7% improvement in network throughput and a 59.5% reduction in packet delay compared to traditional static TDMA. Compared to the naïve Dynamic-Distributed TDMA, D2TRA-RM provides a 125.7% increase in throughput and 68.7% lower delay while keeping the MAC overhead manageable. These results highlight the algorithm’ s ability to effi- ciently allocate resources in highly dynamic environ- ments, significantly boosting overall network perform- ance while maintaining minimal control overhead.

The remainder of this paper is organized as follows. Section II reviews related work and discusses their limitations in the context of robotic MANETs. Section III details the proposed dynamic distributed TDMA resource allocation algorithm. Section IV presents the simulation setup and performance evaluation results. Finally, Section V concludes the paper and suggests directions for future research.

Ⅱ. Background

The development of resource allocation algorithms for MANETs has been an active area of research for several decades^[2,3,6,7]. These networks, which operate without fixed infrastructure, present significant chal- lenges due to their dynamic nature, limited bandwidth, and the need for efficient communication protocols^[8]. In the context of robotic MANETs, these challenges are further exacerbated by the high mobility of the nodes, leading to frequent topology changes, network partitioning, and merging.

Early works on resource allocation for MANETs primarily focused on centralized algorithms, where a central controller would manage the allocation of communication resources among the nodes. Centralized TDMA algorithms^[9] were proposed to handle bandwidth constraints by dividing the available resources among nodes in a coordinated manner. However, these centralized approaches are unsuitable for robotic MANETs due to the inherent scalability issues and the single point of failure problem. In high- ly dynamic environments, such as those involving ro- botic nodes, centralized control can lead to delays in resource allocation and inefficient use of the available bandwidth, especially when the network is partitioned.

To address these limitations, decentralized and dis- tributed algorithms have been developed, allowing no- des to coordinate resource allocation with their neigh- bors autonomously. Distributed TDMA-based proto- cols (e.g., DRAND^[10] and ADCA^[11]) enable nodes to negotiate time slots locally, reducing the need for a central controller. However, these protocols generally assume static or slowly changing network topologies, making them less effective in the highly mobile envi- ronments of robotic MANETs. In scenarios where the network topology changes frequently, these protocols may experience high control overhead and increased communication delays as nodes must constantly re- negotiate resources.

In response to the dynamic nature of robotic MANETs, several dynamic resource allocation algo- rithms have been proposed. For example, the dynamic slot assignment (DSA)^[12] algorithm adjusts time slots dynamically based on traffic conditions. Similarly, mobility-tolerant TDMA protocol (M-TDMA)^[5] takes node mobility into account when scheduling time slots, reducing collisions and improving overall net- work performance.

Despite these advancements, current resource allo- cation algorithms for MANETs still face significant challenges when applied to robotic networks. The high mobility of robots, frequent changes in network top- ology, and the need for efficient partitioned operation have not been fully addressed. Moreover, existing sol- utions often fail to consider the dynamic and time-varying nature of communication demands in ro- botic networks, leading to suboptimal resource uti- lization and increased communication delays. The limitations of the existing resource allocation algo- rithm necessitate a decentralized approach tailored to robotic MANETs.

Ⅲ. Design

The dynamic and distributed nature of robotic MANETs introduces unique challenges that traditional TDMA protocols often fail to accommodate, including rapid topology changes, varying traffic demands, and the need for distributed control in robotic networks. To overcome these limitations, we propose D2TRA-RM, a decentralized, demand-based resource allocation algorithm specifically designed for robotic MANETs. D2TRA-RM enables each node to in- dependently coordinate resource allocation with its neighbors, ensuring efficient operation in both merged and partitioned network scenarios. By dynamically ad- justing resources based on real-time communication demands, optimizing the TDMA frame structure, and minimizing overhead, D2TRA-RM effectively ad- dresses mobility, network partitioning, and low-la- tency requirements while maximizing bandwidth efficiency.

3.1 Core System Architecture

This section describes the dedicated frame and slot structure required for the operation of D2TRA-RM, as well as the fundamental structure of its distributed dynamic resource allocation messages and slot state machine.

Slot Structure and TDMA Frame Design The slot structure, which serves as the basic unit of TDMA resource allocation in the proposed system, is de- signed to efficiently handle both control and data com- munication while ensuring accurate synchronization and gain control, drawing inspiration from prior stud- ies^[13,14]. Each slot consists of three key fields: an AGC field, a PRMB field, and 8 OFDM symbols. The AGC field configures the receiver gain for the upcoming data in the slot, ensuring optimal signal reception. The PRMB field facilitates time and frequency synchroni- zation, which is critical for maintaining coherent com- munication across the network. Additionally, there is a mute time between consecutive slots to enhance syn- chronization further and prevent interference.

The proposed TDMA frame structure divides the frame into fixed slots and dynamic slots, providing a flexible communication framework. While previous studies^[15,17] typically use a single, shared control peri- od that focus on solely for control purposes, our ap- proach employs per-node fixed slots that are evenly distributed across the frame. This ensures fair access to control resources and allows nodes to prioritize control messages, such as HELLO messages for neighbor discovery and network synchronization, in any slot they own. Additionally, fixed slots can also flexibly handle data transmissions when resources are available, supporting critical mission-specific data transmissions without prior negotiation. In parallel, dynamic slots are allocated to nodes based on re- al-time traffic demands via negotiation similar to prior dynamic TDMA works^[15,17]. This ensures efficient re- source utilization and enables the system to adapt ef- fectively to varying network conditions.

Fig. 2.

TDMA frame structure with fixed and dynamic slots.

Furthermore, D2TRA-RM allows for a diverse ratio between fixed and dynamic slots. In scenarios with low traffic and a need for stability, more fixed slots can be allocated to ensure consistent control signaling. During periods of high traffic, dynamic slots are pri- oritized to maximize data throughput. Figure 2 illus- trates the detailed structure of the proposed TDMA frame, including the slot composition and the mute time intervals between slots.

Periodic HELLO Messages For Neighbor Resource Information Sharing D2TRA-RM employs periodic HELLO messages to ensure nodes maintain an up-to-date view of neighboring slot allocations. Each node broadcasts a HELLO message at regular inter- vals (typically every second) to all neighboring nodes. These messages contain a bitmap of the slots currently held by the node and the aggregated information about slot ownership from 2-hop neighbors. This allows each node to maintain a comprehensive view of re- source usage across the local network (Figure 3, 4).

Fig. 3.

Diagram of Hello message exchange between nodes.

Fig. 4.

Slot bitmap example (100 slots).

The periodic sharing of this information ensures that neighboring nodes are aware of available and oc- cupied slots, preventing slot conflicts. Moreover, if discrepancies arise between a node’s view of slot us- age and the HELLO messages it receives, the node can correct its slot allocation map, minimizing the chance of resource contention. While prior studies of- ten rely on more detailed bitmaps to exchange 2-hop resource usage information^[15], our approach opts for a simpler binary representation to indicate slot avail- ability within 1-hop. This design naturally accom- modates 2-hop considerations based on the bitmaps of neighboring nodes, potentially reducing overhead. This streamlined HELLO message exchange helps maintain synchronized resource allocation across no- des within the communication range.

Slot-Specific State Machine Design To manage the dynamic allocation of slots, each node in D2TRA-RM operates a state machine for each individual slot. The state machine defines the current status of each slot and governs the transitions based on control messages received or sent. The primary states for each slot are as follows (Figure 5):

· FREE: The slot is unallocated and is available for use.

· REQ_SENT: The node has sent a request for the slot and is awaiting approval from neighbors.

· GRANTED: The slot has been granted by all rele- vant neighbors and is awaiting use.

· ALLOCATED: The slot is actively used by the node or has been allocated for future use.

· RELEASED: The slot has been released and is tran- sitioning back to the FREE State.

Fig. 5.

Slot-specific state machine visualization.

Each slot’s state machine allows D2TRA-RM to efficiently manage slot transitions, ensuring that each slot is either in use or readily available. By employing state machines for each slot, the system minimizes underutilization and improves slot negotiation efficiency.

3.2 Resource Management Mechanisms

Dedicated Control Messages For Resource Allocation D2TRA-RM uses several dedicated con- trol messages to facilitate the resource allocation process. These messages ensure efficient communica- tion between nodes when negotiating, granting, or re- leasing slots. The main types of control messages used for resource allocation are shown in Table 1. These messages are designed to be lightweight, reducing the overhead of resource negotiation and enabling effi- cient communication even in highly dynamic networks. Each message is tailored to minimize un- necessary network traffic while ensuring reliable slot allocation.

Distributed Slot Negotiation Process The core of the algorithm is the distributed slot negotiation process, which allows nodes to autonomously request or release communication slots based on their current traffic needs. This decentralized approach avoids the need for a central coordinator and reduces the overhead typically associated with centralized resource allocation schemes.

When a node experiences increased traffic and needs more slots, it initiates a slot request procedure. The node broadcasts a REQUEST message to its neighboring nodes, specifying the number of slots it needs and providing its current traffic load and slot usage information. Neighboring nodes evaluate the re- quest by checking their slot usage and the usage of their 2-hop neighbors. Neighbors respond with a GRANT message if the requested slots are available. Otherwise, they return a REJECT message, prompting the requesting node to retry with different slots or de- lay its request.

Algorithm 1 outlines this slot request procedure in detail.

Algorithm 1

Slot Request Procedure

Similarly, when a node no longer needs its allo- cated slots, it releases them by sending its neighbors a RELEASE message. This ensures that the communi- cation slots are efficiently utilized across the network and that idle slots are available to nodes with higher traffic demands. The distributed nature of this process allows each node to make slot allocation decisions in- dependently without requiring global synchronization, which is crucial in mobile and decentralized environ- ments like robotic MANETs.

Handling Topology Changes One of the key chal- lenges in robotic MANETs is the frequent changes in network topology due to node mobility. The algo- rithm addresses this issue by dynamically adapting to topology changes without requiring global coordination. Each node periodically broadcasts HELLO messages that contain a bitmap of the node’s current slot usage. Neighboring nodes use this in- formation to update their local views of the network, ensuring that slot allocations remain conflict-free. When a node detects that one or more neighbors are no longer reachable (indicated by the absence of HELLO messages over a certain period), it initiates a reallocation process to free up any slots previously shared with those neighbors. This process ensures that the network remains flexible and responsive to changes in connectivity, preventing communication breakdowns in highly mobile environments.

3.3 Dynamic Slot Allocation Algorithm Details

This section details the core mechanisms and de- sign principles of the D2TRA-RM algorithm for re- source allocation in highly dynamic robotic MANETs. By addressing key challenges such as traffic fluctua- tions, topology changes, and latency requirements, the algorithm ensures efficient and reliable communication.

3.3.1 Dynamic Slot Request and Release Strategy

In robotic MANETs, some nodes-particularly those serving as relay points or tasked with more complex operations-often experience higher traffic loads. These high-traffic scenarios occur when specific nodes han- dle a disproportionately high number of communica- tions due to their role in the network topology or ap- plication requirements. This leads to two key chal- lenges: (1) slot contention, where multiple nodes com- pete for the same communication slots, and (2) re- source shortages, where high-traffic nodes fail to ac- quire sufficient slots to meet their communication demands. As a result, the network’s overall efficiency is reduced, and packet loss becomes more frequent, particularly in dense or high-load environments. D2TRA-RM addresses these challenges through be- low complementary mechanisms.

Adaptive Slot Request Interval Each node dynam - ically adjusts its slot request interval based on traffi - crelated metrics, such as transmission queue length, dropped packets, and the number of neighboring nodes. Specifically, nodes with longer transmission queues and more dropped packets shorten their re- quest intervals, making them more likely to obtain ad- ditional slots and avoid congestion. This dynamic ad- justment is designed to reduce resource shortages for heavily burdened nodes, directly addressing the dis- parity in resource distribution caused by traffic imbalances.

The interval is calculated as follows:

Equation 1 prioritizes nodes experiencing high com- munication demands, ensuring they receive additional slots more quickly. By integrating both traffic-related metrics (queue length and packet drops) and network density (number of neighbors), the model dynamically balances resource allocation among nodes. Nodes with longer queues need slots more urgently, so the interval between slot requests is shortened to ensure faster re- source allocation. The dropped packets parameter re- flects the number of packets lost due to insufficient slot allocation; nodes experiencing packet loss also decrease their request interval to prioritize recovering from the slot shortage. Finally, the number of neigh- boring nodes is included to avoid overwhelming the network with excessive slot requests in denser environments. Nodes with more neighbors slightly in- crease their request interval to balance the overall load on the network.

For example, in cases where a node has a long transmission queue and has dropped several packets, the slot request interval is reduced significantly, al- lowing the node to request slots more frequently. This ensures that nodes under high traffic load or suffering from packet drops are given priority when competing for available slots. On the other hand, nodes with shorter transmission queues and fewer dropped pack- ets maintain a longer interval between slot requests, reflecting their lower priority for additional resources.

Fig. 6.

Slot request interval adjustment based on traffic load and packet drops.

The following diagram illustrates two scenarios (Figure 6). In Case 1, a node with a long transmission queue and multiple dropped packets requests slots more frequently, resulting in higher priority for re- source allocation. In Case 2, a node with fewer com- munication demands has a longer interval between re- quests, reducing its priority for additional slots. This adaptive slot request mechanism ensures that resource allocation is dynamically adjusted to match node-level traffic demands, reducing the impact of slot contention and resource shortages.

Prime Number-Based Slot Request To further re- duce contention, D2TRA-RM employs a primenum - ber-based slot allocation strategy. Each node is as- signed a unique prime number and attempts to allocate slots based on multiples of that number. For example, a node with the prime number 13 will attempt to allo- cate slots 13, 26, 39, and so on. This staggered alloca- tion method effectively spreads slot requests across the network, reducing the likelihood of collisions and ensuring efficient resource distribution.

Probability-Based Slot Release and Retention Traffic patterns fluctuate rapidly in highly dynamic robotic networks based on nodes’ tasks and movements. Immediate slot release after traffic drops can lead to unnecessary reallocation cycles when traf- fic increases shortly after. In contrast, holding unused slots for too long reduces the available resources for other nodes.

D2TRA-RM addresses this problem with a proba - bility-based slot release mechanism. Instead of releas- ing a slot once its utilization drops to zero, the system retains it with a probability of 50%. This approach balances the need to free up resources with the poten- tial for short-term traffic fluctuations. By holding onto slots that may be needed again shortly, D2TRA-RM reduces the overhead of frequent slot reallocations, optimizing overall network performance while mini- mizing unnecessary control traffic. Additionally, D2TRA-RM incorporates a slotretentionstrategythat ensures low-utilization slots are not immediately released. This strategy prevents excessive control overhead from slot release and reallocation processes, ensuring the system maintains a stable communication flow even during fluctuating traffic.

3.3.2 Relay Node Resource Management

In robotic MANETs, multi-hop (e.g., 5) communi- cation is often required to reach nodes across large distances. Relay nodes along these paths can experi- ence resource shortages, especially in high-traffic environments. Moreover, time-sensitive traffic, such as control commands or safety-critical data, requires guaranteed low-latency transmission, which is diffi- cult to achieve under typical slot allocation processes.

To address resource shortages at relay nodes, D2TRA-RM implements a slot-stealingmechanism . When a relay node detects insufficient slots to forward packets, it sends a targeted resource allocation request to its predecessor node. This request is a dedicated control message called REQUEST, which contains the address of the relay node and the specific slots it wish- es to steal from its predecessor. The predecessor node responds with a GRANT message for the requested slots, allowing the relay node to immediately use them for forwarding traffic.

This mechanism differs from the standard slot allo- cation process in D2TRA-RM, where slot requests are broadcast to all neighboring nodes, and the requesting node must receive a GRANT from all neighbors be- fore using the slot. In the slot-stealing process, the relay node only communicates with its predecessor to expedite the resource transfer. This targeted approach significantly reduces the delay in securing additional slots, ensuring that relay nodes can quickly adapt to traffic surges.

For example, consider a scenario where Node 2, acting as a relay, receives packets from Node 1 (Figure 7). If Node 2’s transmission queue contains eight slots worth of data from Node 1 but lacks the necessary slots to transmit, it sends a REQUEST to Node 1, specifying four slots it wishes to steal (half of the required resources). Upon receiving the GRANT from Node 1, Node 2 immediately uses the stolen slots to forward the packets, ensuring smooth multi-hop communication without delay. By effi- ciently reallocating resources between relay nodes and their predecessors, the slot-stealing mechanism pre- vents resource bottlenecks and maintains the flow of multi-hop transmissions.

Fig. 7.

Diagram of REQUEST message and slot-stealing mechanism for relay nodes.

3.3.3 Slot Preemption for Critical Traffic

In time-sensitive applications on robotics MANET, such as control commands or safety-critical data trans- mission, D2TRA-RM implements a slot preemption algorithm to guarantee low-latency communication for specific high-priority traffic. This algorithm guaran- tees that critical packets, such as control commands or emergency data, can traverse a 5-hop path with a total delay of less than a certain bar (e.g., 50ms). When critical traffic such as control commands or emergency data is detected, the slot preemption algo- rithm is activated.

The slot preemption algorithm identifies pre- emptible slots along the 5-hop path from the source node S to the destination node T. The preemption in- terval (S_int) is set to 4 slots, ensuring that each hop introduces no more than a 4ms delay (D_hop). This de- sign ensures that the critical packet is transmitted across the entire 5-hop path with a maximum end-to-end latency (L_max) of 50ms.

Algorithm 2

Slot Preemption Algorithm for Critical Traffic

Once preemptible slots are identified at each node, any ongoing non-critical traffic using those slots is preempted, and the critical packet is allocated the pre- empted slots for transmission. This guarantees that the critical traffic bypasses any lower-priority traffic and meets the strict latency requirements for time-sensitive applications.

This slot preemption mechanism is essential for en- suring that real-time communications, such as control commands or safety-critical data, are transmitted with- out delay, even under high network load. By enforcing the strict delay constraint, D2TRA-RM provides a re- liable solution for mission-critical communication in robotic networks.

Fig. 8.

Example of slot preemption for critical 5-hop traffic with less than pre-defined delay.

3.3.4 Control Overhead Reduction

In large-scale robotic MANETs, where many nodes compete for limited resources, the overhead from con- trol messages used for slot allocation and release can become significant. Due to packet loss or network congestion, frequent communication failures further exacerbate this problem by causing repeated allocation attempts.

To reduce control overhead, D2TRA-RM imple- ments a probability-based forced slot alloca - tion-releasemechanism . When a node fails to receive confirmation from all its neighbors after sending a slot request, it proceeds with the slot allocation with a 20% probability. This reduces the overhead from repeated allocation attempts, especially in high-traffic scenarios where control messages might be lost. If a slot conflict arises, the system uses collision resolution mecha- nisms to reassign slots, reducing the need for con- tinuous control message exchanges and improving throughput.

3.3.5 Integrated Design Outcomes

The D2TRA-RM algorithm seamlessly integrates multiple adaptive strategies to address the challenges of resource allocation in robotic MANETs. By com- bining dynamic slot request intervals, prime num- ber-based allocation, probability-based slot retention, and preemption mechanisms, the system ensures effi- cient resource utilization, low-latency communication for critical data, and reduced control overhead. These mechanisms collectively adapt to dynamic traffic de- mands, minimize slot contention, and support seam- less multihop communication, enabling real-time, reli- able performance in highly dynamic and large-scale networks.

Ⅳ. Performance Evaluation

In this section, we evaluate the performance of the proposed D2TRA-RM algorithm through simulations. The evaluation focuses on measuring the overall effi- ciency of the system under different network con- ditions, particularly in terms of solving the problems outlined in the previous section. The results demon- strate the effectiveness of D2TRA-RM in managing resources, reducing communication latency, and adapting to network topology changes in robotic MANETs.

4.1 Simulation Setup

The simulated network consists of 31 mobile nodes structured to emulate practical robotic MANET con- figurations such as those used in swarm robotics or emergency response operations. The network includes one central operational device, five clusters of five nodes, each led by a leader, and five intermediary no- des positioned to facilitate communication between the clusters and the operational device. This layout mimics realworld scenarios involving collaborative robotic systems or coordinated response teams. The topology and node arrangement are illustrated in Figure 9.

The simulation runs for a total of 100 seconds and is divided into three key phases:

1. Deployment phase (0-10s): Nodes are initially deployed and move to predefined positions within the simulation area.

2. Communication initialization phase (10-20s): Task Coordination (TC), Task Management (TM), and data transmissions begin across the network.

3. Operational phase (20-100s): Nodes engage in random movement within assigned regions, with a maximum speed of 1 m/s. During this phase, every 20 seconds, a randomly selected node ini- tiates a video upload to simulate high-bandwidth traffic.

Fig. 9.

Network topology and traffic patterns in the simulation scenario.

The movement speed of 1 m/s was selected as it aligns with the typical mobility patterns observed in swarm robotics and related MANET simulations, en- suring that the evaluation accurately reflects practical movement behaviors^[18,19].

The simulation scenario incorporates four distinct types of traffic, representative of the diverse commu- nication needs found in typical robotic MANET appli- cations:

· Task Coordination (TC): 30 kbps data stream from the operational device to all other nodes, facilitating the coordination of tasks.

· Task Management (TM): 30 kbps traffic from the leader nodes to the operational device, handling task-related communication.

· Video Traffic: High-bandwidth 1.5 Mbps video data sent from follower nodes to the operational device, simulating the upload of video feeds.

· Operational Data: 100 kbps of operational data transmitted from follower nodes to their re- spective leader nodes.

This variety in traffic types is intended to test the performance of D2TRA-RM under different commu- nication loads, reflecting the typical demands of ro- botic networks, particularly in real-time task manage- ment and data-intensive applications.

The simulation takes place in a 100m x 100m envi- ronment, which includes physical obstacles modeled using a free space path loss model^[20]. These obstacles introduce an additional 10% attenuation, simulating walls or barriers commonly encountered in realworld operational settings. Each node has a communication range of 30 meters, sufficient to maintain contact with nearby nodes while accommodating potential gaps due to mobility.

The key parameters used in the simulation are sum- marized in Table 2. The selected parameters represent typical operational conditions for robotic MANETs in applications such as search-and-rescue missions, tac- tical deployments, and coordinated robotic operations. In addition, the routing protocol follows the basic RFC^[21] specification. D2TRA-RM is evaluated under these conditions using a dynamic TDMA protocol that allows deterministic slot assignment, ensuring colli- sion-free communication in highly dynamic environments. This deterministic behavior is crucial for maintaining reliable and predictable communica- tion in mission-critical operations, where both throughput and low-latency delivery are essential for effective coordination.

4.2 Performance Metrics

To evaluate the performance of the D2TRA-RM system, we consider three key metrics: Throughput (kbps), MAC Overhead (kbps), and Delay (msec). These metrics provide insights into how efficiently the system handles resource allocation, particularly under varying traffic demands and network conditions.

Thr oughput (kbps) Throughput is measured in kilo-bits per second (kbps) and reflects the total amount of data successfully transmitted across the net- work over time. Higher throughput indicates better network efficiency, which is critical in supporting high-bandwidth applications like video streaming or data-intensive sensor communications in robotic networks.

MAC Over head (kbps) MAC overhead refers to the control message overhead introduced by the Medium Access Control (MAC) layer. It is measured in kilo-bits per second (kbps). Higher MAC overhead suggests that more bandwidth is being used to manage slot allocation, synchronization, and other network op- erations, potentially reducing the bandwidth available for actual data transmission. Lower MAC overhead indicates more efficient resource management.

Delay (msec) Delay is the time required for a packet to travel from its source to its destination, measured in milliseconds (msec). Lower delay is essential for realtime applications in robotic systems, such as con- trol commands or time-sensitive data transmissions. Minimizing delay ensures that critical data reaches its destination quickly, which is important for maintain- ing coordination between nodes in a robotic MANET.

These performance metrics provide a compre- hensive understanding of the D2TRA-RM system’s effectiveness in optimizing resource allocation, re- ducing communication delays, and managing network overhead, particularly in dynamic and decentralized environments.

4.3 Results and Analysis

The experimental results validate the effectiveness of the D2TRA-RM system across various network configurations and conditions. This section analyzes the results for each problem-solution pair, using throughput, MAC overhead, and delay as key per- formance indicators.

4.3.1 Analysis on Proposed Dynamic TDMA Frame Structure

The first set of experiments compared network per- formance when using a fully fixed slot allocation (100:0) against a mixed fixed-dynamic slot allocation (60:40). Bold values indicate the best performance, while underlined values show the second-best. As shown in the results, the dynamic allocation provided a significant increase in throughput, from 1008 kbps to 1334 kbps, while introducing a moderate increase in MAC overhead (from 0 to 156 kbps) and delay (from 672 ms to 869 ms). These results illustrate the trade-offs between throughput and control overhead when dynamic slot allocation is used to better adapt to varying traffic demands. It is important to note that the reason for the lower delay observed in the fully fixed slot allocation (100:0) is that dropped packets are not included in the delay measurement. This means that, while the delay is lower, the lower throughput negatively impacts overall performance, as fewer packets are successfully transmitted. The lower throughput in the fully fixed slot allocation indicates that the system is less capable of handling varying traffic demands, leading to packet loss and lower overall performance.

The increased throughput observed in the dynamic scenario is primarily due to the system’s ability to allocate resources dynamically based on traffic re- quirements, preventing packet loss for critical applica- tions such as video streaming. However, the increase in MAC overhead indicates the cost of managing this dynamic allocation process.

Proportional Slot Allocation for Different Traffic Loads Further experiments explored the ef- fects of varying the fixed-to-dynamic slot ratio (e.g., 30:70, 40:60, and 60:40). As depicted in Figure 10 and Table 3, the 60:40 ratio achieved the best balance, with the highest throughput of 1334 kbps, while main- taining acceptable MAC overhead (156 kbps) and de- lay (869 ms). The 30:70 and 40:60 configurations also performed well but with slightly lower throughput and higher overhead. These results show that adjusting the slot ratio based on network traffic can significantly improve performance by allocating more dynamic slots to handle peak loads without overly increasing overhead.

Fig. 10.

Performance (throughput and delay) visualization for Various fixed-to-dynamic slot ratios. The translucent areas on the graph represent the variance when averaging over 10 random seeds.

4.3.2 Ablation Studies on Algorithm Details

Adaptive Slot Request Intervals To address the problem of resource imbalance, the D2TRA-RM sys- tem adapts the slot request interval based on node traf- fic demands (Equation 1). Detailed evaluation results are in Table 4. When the interval is dynamically ad- justed according to the node’s transmission queue length and dropped packets, throughput improved from 2429 kbps to 3001 kbps, while MAC overhead dropped from 665 kbps to 374 kbps. The delay re- mained relatively stable, demonstrating that adjusting the request interval leads to more efficient resource usage without significantly affecting latency. The re- sults highlight that this mechanism resolves resource imbalances by ensuring nodes with higher traffic de- mands request slots more frequently, preventing de- lays and minimizing packet loss.

Probability-based Slot Allocation The system was evaluated for its ability to allocate slots probabilisti- cally when acknowledgments (ACKs) were not re- ceived from neighboring nodes. Specifically, we test- ed different slot allocation probabilities (1/3, 1/4, 1/5, and 1/6). The results demonstrated that the 1/5 proba- bility yielded the best balance of throughput and de- lay, significantly improving network performance compared to the strict slot allocation scenario (0 prob- ability).

The results showed that throughput improved from 1772 kbps to 2941 kbps with a 1/5 probability while keeping the delay to 273 ms. This mechanism is par- ticularly advantageous in environments with high packet loss, as shown in the packet loss scenarios evaluated.

Probability-based Slot Release Another experiment tested the effectiveness of probabilistically retaining slots after their utilization became 0. Instead of releas- ing slots immediately, we introduced a mechanism that kept them with a certain probability (1/3, 1/2, or 1/3). This mechanism aimed to reduce unnecessary reallocation delays and slot overhead. The simulation result is in Table 6.

The 50% probability setting yielded the best per- formance with throughput peaking at 3011 kbps, while both MAC overhead (274 kbps) and delay (272 ms) were minimized. This shows that retaining slots for a brief period, even after usage, optimizes resource utilization without causing significant overhead.

Relay Node Resour ce Shor tage Handling The relay node resource shortage issue was examined by simu- lating a scenario where nodes needed to relay traffic but had insufficient slots. The system dynamically re- allocated slots by preempting resources from upstream nodes. The visualized simulation demonstrated the ef- ficiency of this approach in ensuring seamless com- munication, even under high-load conditions. As in Figure 11, the resource preemption and redistribution algorithm successfully alleviated the resource bottle- neck issue for multi-hop relay nodes. Details are the following: At 12 seconds, Node A generated video traffic, exhausting its resources and preventing Node B from forwarding packets to Node C. By 13 seconds, Node B preempted half of Node A’s slots, resuming the relay. This process continued with Node C taking resources from Node B. By 15 seconds, the resource allocation stabilized, ensuring proper relay operation along the chain.

Fig. 11.

Simulation of resource reallocation for relay nodes (time 12s to 15s). The green line represents packet throughput, and the red arrows indicate packet drops.

Slot Preemption for Critical Traffic In military and mission-critical scenarios where low-latency trans- mission is essential, the slot preemption algorithm is vital in ensuring that critical traffic is transmitted across a multi-hop path with an end-to-end latency of less than a certain value. We evaluated our algo- rithm on the scenario that consists of 5-hop with a 50ms latency bar. The evaluation shows that the max- imum delay was reduced from 872 ms (without pre- emption) to 169 ms, with an average delay reduction from 365 ms to 64 ms(Table 7). This mechanism en- sures that high-priority packets are given precedence over non-critical traffic, which is especially important in time-sensitive applications. As illustrated in Figure 12, the delay observed at around 20 seconds was caused by nodes sequentially receiving slot allocations. The red circle highlights this occurrence, showing how the preemption algorithm helped reduce the latency. After that circle, the lower delay is con- sistent for a while. This improvement confirms the preemption algorithm’s role in ensuring low-latency communication, especially for critical data such as control commands.

Fig. 12.

The preemption algorithm ensures low-latency transmission for critical traffic.

4.3.3 Overall Performance of D2TRA-RM

The final experiment compared the performance of D2TRA-RM against two baseline schemes: a tradi- tional TDMA (with all nodes equally allocated slots) and a naïve dynamic-distributed TDMA (with a 60:40 fixed-to-dynamic slot ratio). The following table sum- marizes the results(Table 8).

D2TRA-RM achieves significant performance gains over both TDMA and naïve Dynamic-Distributed TDMA schemes. In terms of throughput, D2TRA-RM delivers 3011 kbps, which represents a 198.7% improvement over TDMA (1008 kbps) and a 125.7% improvement over naïve Dynamic-Distributed TDMA (1334 kbps). These re- sults highlight the system’s ability to allocate re- sources dynamically, ensuring that high-demand no- des receive the necessary bandwidth

Regarding delay, D2TRA-RM reduces latency by 59.5% compared to TDMA and by 68.7% compared to the naïve Dynamic-Distributed TDMA, ensuring that time-sensitive traffic is delivered more efficiently. While MAC overhead increases from 0 in the TDMA scheme to 274 kbps in D2TRA-RM, this is a small trade-off considering the substantial gains in through- put and latency.

These results demonstrate that D2TRA-RM handles dynamic traffic and resource reallocation more effi- ciently and far outperforms traditional and naïve dy- namic TDMA approaches, making it a highly suitable solution for complex and dynamic robotic MANET environments.

Ⅴ. Discussion and Conclusion

The proposed D2TRA-RM algorithm effectively addresses key challenges in robotic MANETs, such as dynamic topology changes, varying traffic de- mands, and decentralized control. Despite certain limi- tations, D2TRA-RM provides a robust framework for efficient resource allocation in these networks. In this section, we discuss the algorithm’ s overall effective- ness, its limitations, potential future improvements, and possible applications.

Effectiveness of D2TRA-RM D2TRA-RM con- sistently demonstrates high slot utilization and adapt- ability, ensuring efficient bandwidth usage even in dy- namic environments. The probabilistic delay mecha- nism successfully reduces slot conflicts in high- mobility scenarios, allowing nodes to manage commu- nication resources without centralized control. The priority handling system ensures that critical traffic, like control commands, experiences minimal delay, which is essential for real-time applications in robotic operations.

Moreover, the algorithm’ s low control overhead allows more bandwidth for data transmission, making it suitable for bandwidth-constrained environments. This efficiency is vital in scenarios such as emergency response or tactical operations, where reliable commu- nication is critical.

Limitations D2TRA-RM demonstrates robust per- formance; however, certain challenges remain. In ex- tremely dense networks, decentralized control may oc- casionally result in slot conflicts when multiple nodes simultaneously request the same slots, despite the ran- dom backoff mechanism. Furthermore, the algo- rithm’s reliance on periodic HELLO messages for top- ology updates may introduce brief inefficiencies dur- ing rapid topology changes or high mobility scenarios. Lastly, while the fixed TDMA frame structure effec- tively balances stability and adaptability, it may limit the system’ s responsiveness under highly variable or bursty traffic conditions.

Future Work To further enhance D2TRA-RM, future research could explore predictive models or machine learning techniques to anticipate and proactively re- solve slot conflicts in dense network environments. For instance, D2TRA-RM could adapt work on ma- chine learning and conflict prediction suggests that ap- propriate machine learning methodologies can offer substantial improvements in accuracy and perform- ance^[22]. Refining the HELLO message mechanism by adopting event-driven or context-aware updates could improve the system’ s adaptability to rapid topology changes. Exsiting dynamic hello messaging scheme for neighbor discovery in on-demand MANET routing protocols indicates that making hello intervals propor- tional to event intervals can reduce unnecessary hello messages, thereby enhancing energy efficiency^[23]. Additionally, incorporating a hybrid TDMA protocol that combines fixed slots for routine tasks with flexi- ble slots for dynamic traffic could enhance scalability and responsiveness under variable traffic loads. The development of such hybrid protocols has been dis- cussed in the context of ad hoc networks, aiming to achieve high channel utilization under varying con- tention levels^[24]. These improvements would strength- en D2TRA-RM’s applicability in mission-critical sce- narios such as urban search and rescue or military op- erations requiring real-time, robust communication.

Potential Applications D2TRA-RM has immediate applications in areas such as disaster response, where autonomous robots need reliable communication to coordinate tasks in real time. Its ability to prioritize critical traffic makes it suitable for environments where timely data transmission is essential, such as in locating survivors or assessing hazards.

In military operations, D2TRA-RM’ s decentral- ized approach and low-latency performance are ad- vantageous for tactical deployments, especially in hos- tile or infrastructure-limited environments. The algo- rithm can also be applied in industrial automation, where coordination between autonomous systems re- quires both efficiency and reliability, ensuring that critical commands are executed without delay.

Conclusion In conclusion, D2TRA-RM provides a flexible and efficient solution for resource allocation in robotic MANETs. It handles the complexities of dynamic environments through adaptive slot alloca- tion, effective prioritization of traffic, and low control overhead. While there are areas for improvement, the overall performance of D2TRA-RM makes it well-suited for real-world applications. Future en- hancements, such as improved conflict resolution and more sophisticated mobility models, will further strengthen its role in advanced robotic communication systems.