LTDA-MAC v2.0: Topology-Aware Unsynchronized Scheduling in Linear Multi-Hop UWA Networksâ€
Round 1
Reviewer 1 Report
In this paper the authors extend their previous works by proposing a greedy optimization algorithm to find the optimal transmission schedule for LTDA-MAC, a MAC layer where spatially separated nodes can perform simultaneous transmissions. This MAC can be used in linear networks. The proposed algorithm outperforms the GSA+PSO solution both in terms of MAC schedule and computational cost.
Compared to other related works, the proposed schedule does not require timing synchronization between the nodes and requires a low computational cost, thus significantly improving the state of the art.
LTDA-MAC is clearly designed for linear networks, and this is well explained in the paper. However the reviewer wonders whether or not LTDA-MAC can also be used in non linear networks.
The LTDA-MAC is evaluated by simulating two linear networks with different propagation delay, hence proving the effectiveness of the proposed solutions in different linear network configurations. The proposed schedule requires a smaller frame than other protocols, like STDMA.
Other related works that might be cited are:
Y. Noh et al., "DOTS: A propagation delay-aware opportunistic MAC
protocol for mobile underwater networks", IEEE Trans. Mobile Comput.,
vol. 13, no. 4, pp. 766–782, Apr. 2014.
R. Diamant et al, "A handshake-based protocol exploiting the near-far effect in underwater acoustic networks", IEEE Wireless Communications Letters, Volume: 5, Issue: 3, June 2016
Author Response
Thank you for your positive comments and useful suggestions of further references. Please see our responses below, with the changes in the paper highlighted in blue.
1) However the reviewer wonders whether or not LTDA-MAC can also be used in non linear networks.
LTDA-MAC was designed specifically for linear networks, so it cannot be used for other types of network topologies. However, the first paragraph of Section 2 refers to other variants of the TDA-MAC protocol that are designed for other, more conventional types of network topologies. To make it clearer to the reader, we added the following text on p.2 line 70.
In this paper we focus on unsynchronized scheduling in LUASNs using the LTDA-MAC protocol [8] – a protocol designed specifically for networks with line topologies.
2) Other related works that might be cited are:
Noh et al., "DOTS: A propagation delay-aware opportunistic MAC protocol for mobile underwater networks", IEEE Trans. Mobile Comput., vol. 13, no. 4, pp. 766–782, Apr. 2014.
Diamant et al, "A handshake-based protocol exploiting the near-far effect in underwater acoustic networks", IEEE Wireless Communications Letters, Volume: 5, Issue: 3, June 2016
Since the suggested references describe contention-based MAC protocols, as opposed to schedule-based protocols that are the focus of this paper, we expanded paragraph 3 of the Introduction to discuss contention-based MAC and include references to the notable examples in the underwater acoustic domain (including the two suggested by the reviewer). The new content added on p.2 l. 36-47 is the following:
As an alternative to deterministic schedule-based TDMA methods, communication networks often use contention-based MAC protocols [15–19] where nodes access a shared channel randomly on demand, based on a particular set of rules [20]. However, most conventional contention-based MAC protocols are highly inefficient in the underwater acoustic environment. For example, channel reservation based protocols waste a large part of channel capacity while the nodes are waiting for control signals to propagate through the slow acoustic medium to establish a communication link, e.g. Request-to-Send (RTS), Clear-to-Send (CTS), acknowledgements etc. These waiting times result in significant loss of throughput and poor channel utilization [13,15,16]. This paper focuses on the schedule-based MAC approach, as it is particularly well-suited for the linear network topologies, where the sparse node connectivity can be exploited to schedule simultaneous, spatially separated transmissions.
Reviewer 2 Report
The paper presents a simple greedy algorithm to derive schedules for linear underwater acoustic networks accessing the MAC channel through TDMA schemes.
The greedy algorithm consists of increasing the transmit delay allocated to some node in a way that during this time there will be no collisions.
The paper is comparatively short for a journal. Yet, the authors explain the problem clearly and provide sufficient context, so I would be inclined to name a couple of revisions that should be made before the paper can be ready for publication.
1) In the results, I find it suspicious that the GA+PSO version of LTDA-MAC finds solutions that are so much longer than those of the greedy protocol in terms of frame duration. Can this be due to the fact that the GA or PSO stages do not explore the solution space with a sufficiently large offspring, in order to keep the level of complexity tractable?
2) Can the greedy algorithm be employed to find a first solution, and then GA+PSO be used to explore the nearabouts of this solution?
Author Response
Thank you for your positive and insightful feedback on the paper. Please see our responses to them below, with the changes in the paper highlighted in blue.
1) In the results, I find it suspicious that the GA+PSO version of LTDA-MAC finds solutions that are so much longer than those of the greedy protocol in terms of frame duration. Can this be due to the fact that the GA or PSO stages do not explore the solution space with a sufficiently large offspring, in order to keep the level of complexity tractable?
Thank you for this interesting comment. Indeed, the solutions of the GA+PSO method are visibly worse than those found by the new greedy algorithm. However, our previous extensive simulation studies of the GA+PSO method indicate that this is due to the inherently discontinuous and high-dimensional nature of the optimization problem, where treating the entire scheduling problem as a “black box” makes it very challenging for a heuristic algorithm to find good solutions. In contrast, the new greedy algorithm deconstructs the problem into a series of much simpler problems, minimizing every transmit delay in isolation, thus resulting in much better performance.
We previously had the same concern regarding the parameters of the GA and PSO algorithms, and have therefore performed extensive empirical studies to find the best configurations for them. We added the following text to Section 3.2 to reflect this.
p.5-6 lines 172-180:
The parameters of the GA and PSO algorithms are:
- GA: population size – 500, mutation rate – 0.1, 80% scattered crossover, 1000 generations limit,
- PSO: swarm size – 500, minimum neighbourhood fraction – 0.1, adaptive inertia range – [0.05, 0.8], 1000 iterations limit.
These parameters were empirically found to produce the best performance in our previous study in [8] by providing a large enough population/swarm to explore the high dimensional solution space, and to enable a sufficient number of GA generations and PSO iterations to find good suboptimal solutions.
p.6-7 lines 195-201:
The significant difference in the optimization performance between the proposed greedy algorithm and the GA+PSO method is due to the inherently discontinuous and high-dimensional nature of the optimization problem, where treating the entire scheduling problem as a “black box” makes it very challenging for a heuristic algorithm such as GA or PSO to find good solutions. In contrast, the new greedy algorithm deconstructs the problem into a series of much simpler problems, minimizing every transmit delay in isolation, thus resulting in much better performance.
2) Can the greedy algorithm be employed to find a first solution, and then GA+PSO be used to explore the nearabouts of this solution?
This is an interesting idea. However, a key benefit of the proposed greedy optimization algorithm is the high computation speed, achieved by deconstructing the high-dimensional optimization problem into a series of simpler and smaller problems (minimizing every delay separately). The disadvantage of then employing a GA+PSO algorithm is the high computational cost for a likely negligible benefit. The greedy algorithm will find a solution where every transmit delay is at its minimum such that it satisfies the zero collision constraint. Therefore, it is highly unlikely that the GA+PSO would find any gradual small-scale improvements, e.g. converge towards a local minimum, without introducing one or more collisions into the schedule. Therefore, to reduce any of these delays further (i.e. to find a better solution), the GA+PSO algorithm would need to sufficiently diverge from the current best solution, to start finding zero collision candidate schedules again. Therefore, we believe that the current method using the greedy optimization algorithm is best both in terms of computational efficiency and the schedule solutions that are found.
