Introduction
Problématique et Motivation
Ad hoc wireless networks offer the advantage of rapid deployment at a low cost, making them an effective solution for various contexts, such as providing coverage in underserved areas, establishing temporary networks for events (like sports and concerts), or supporting civil security operations in disaster zones The concept involves deploying a series of nodes/routers equipped with Wi-Fi cards configured in ad hoc mode, which collectively cover the area where users are located Data from these users is transmitted through multi-hop connections across these nodes, with some nodes connected to the Internet, acting as gateways for the rest of the network This infrastructure enables wireless interconnection among users and facilitates access to wired infrastructure, such as Internet connectivity.
A significant issue persists regarding the capacity of networks, defined as the number of bits per second that can be transported from sources to destinations When all nodes utilize the same Wi-Fi channel, network capacity is notably limited due to the sharing of the channel among numerous wireless links, leading to a division of usable bandwidth A simple and cost-effective solution involves equipping nodes with multiple wireless cards, each associated with different channels This approach reduces the number of links using the same channel, resulting in a substantial increase in network capacity However, the assignment of channels to the various radios must be automated and optimized to enhance overall network capacity.
Dans ce travail nous étudions les algorithmes d’assignation de canaux et l’impact du sens du trafic dans la présence des conflits
My work focused on two key areas: the analysis of existing channel assignment algorithms and the evaluation of gains achieved by considering the direction of links (traffic flow) in conflict calculations.
Contribution
Dans ce travail j’ai réalisé deux principales tâches :
The initial step involved implementing the algorithm proposed in [7], which included support for channel allocation based on traffic direction This approach aimed to assess the difference in the number of conflicts between unidirectional and bidirectional traffic, using the metrics outlined in [7].
The second approach involves proposing a mathematical model that considers distances, rather than the number of jumps, to identify conflicting links, with nodes distributed randomly in space.
Plan du rapport
The document is structured as follows: Chapter 2 presents the state of the art and key concepts related to the topic Chapter 3 discusses initial experiments aimed at assessing the impact of traffic direction on the occurrence of conflicts Chapter 4 introduces a mathematical model that evaluates the effect of traffic direction, utilizing probabilistic concepts and distances between nodes Finally, Chapter 5 concludes with our findings and future perspectives.
Concepts et état de l'art
Définition et explication des concepts
A radio link between two nodes exists when both nodes share a common channel and meet specific radio conditions, such as distance and signal quality, to enable effective communication.
Nous disons qu'il y a conflits entre des liens lorsque ces liens ne peuvent pas être utilisés en même temps
There are two types of conflicts in communication systems: conflicts between two transmitters using the same frequency, where each detects the other's signal, and conflicts between a transmitter and a receiver when the transmitter interferes with an illegitimate receiver The authors in [1] defined interferences based on two models: the Protocol Model and the Physical Model [9].
Protocol Model : deux liens e1(u1,v1) et e2(u2,v2) interfèrent entre eux s’ils utilisent le même canal et si toutes les distances d(u1,u2) , d(u1,v1) , d(v1,u2) , d(v1,v2) sont inférieures à Ri (Ri : rayon d’interférence)
A successful transmission occurs when the signal-to-noise ratio (SNRij) exceeds the threshold (SNRthres) Here, SNRij represents the signal-to-noise ratio at node nj for the transmission received from node ni.
Another model based on the number of jumps, known as the Graph model, indicates that two links are in conflict if the number of jumps separating them is less than one, typically two or three.
Two edges e1(u1, v1) and e2(u2, v2) are considered to be in conflict if the minimum of the distances, measured in hops, between their endpoints is less than or equal to a specified threshold SH Here, dH(., ) represents the hop distance between two vertices, indicating the number of hops required to traverse from one vertex to another, while SH denotes the maximum allowable number of hops for conflict determination.
In this study, we examine two models based on a graph framework that considers traffic flow on links We opted for the graph model due to its realistic representation, as distance-based models tend to be overly simplistic In practice, the quality and range of radio links are influenced by geographical factors such as buildings, obstacles, and trees, rather than merely the distances between nodes Furthermore, accurate distance information necessitates a geolocation system, like GPS, which is not always accessible.
The physical model is the most realistic representation; however, information regarding the SINR is not always accessible at the node This metric fluctuates over time due to varying interference levels.
Le graph model a l’avantage de ne requérir que des informations topologiques disponibles grâce à un protocole de routage par exemple De plus, la topologie varie à des échelles de temps raisonnables
A directional link is defined as a connection used in a single direction, where one node transmits data packets to another node without the possibility of reverse communication In the context of this report, a directional link refers to a connection that can physically transmit data only in one direction.
Au contraire des liens directionnels, un lien est bidirectionnel lorsque il est utilisé dans les deux sens ; c'est dire les deux nœuds l’utilisent pour envoyer des paquets de données
Figure 1: Lien directionnel et lien bidirectionnel
Dans ce rapport nous utilisons les expressions ô conflits directionnel ằ et ô conflit bidirectionnel ằ lorsque le conflit concerne un de ces deux types de liens.
les problèmes d’allocation
To enhance network capacity, each node is equipped with multiple wireless cards, allowing for the use of orthogonal channels that can operate simultaneously without interference This approach minimizes conflicts, enabling nearby links to function concurrently without sharing, which ultimately increases throughput.
Due to the limited number of orthogonal channels provided by the IEEE 802.11b/g (3 channels) and IEEE 802.11a (12 channels) standards, it is essential to assign these channels strategically to optimize network capacity.
Network capacity refers to the volume of data that a network can transport Calculating this capacity for a specific assignment can be complex, often leading to directional and bidirectional links that aim to minimize conflicts In the following section, we will present examples of approaches that utilize this principle.
Quelques algorithmes d'assignation de fréquences
Dans [3] ils ont classés les algorithmes selon 3 catégories :
Les fixes : alloue des canaux aux interfaces de faỗon permanente ou pour des intervalles de temps très longs
Les dynamiques : permettent toute interface à être attribuée tout canal, et ces interfaces peuvent souvent passer d'un canal à l'autre [3][9]
Les hybrides : combinent les deux propriétés d'affectation statiques et dynamiques en appliquant une affectation fixe pour certaines interfaces et une affectation dynamique pour d'autres interfaces [3]
Figure 2: classification des algorithmes (source : [3])
MesTiC is a centralized, static greedy algorithm for channel allocation that operates in polynomial time It utilizes a ranking function, which assesses each node based on traffic characteristics, topological properties (specifically the number of hops from the node to the gateway), and the number of radio interfaces per node This ranking system enables the algorithm to visit each node once, thereby avoiding unnecessary returns that could extend execution time Topological connectivity is maintained through a default common channel deployed on a separate radio at each node, which can also facilitate network management.
The central concept behind MesTiC is to assign channels to the radios of a node based on a ranking system that prioritizes nodes The node rank, or Rank(node), determines its priority in channel allocation for incoming links This ranking incorporates the dynamics of channel assignment and is calculated based on three key factors.
Le trafic total d'un noeud basé sur sur la charge offerte du réseau maillé comme calculé dans
La distance du noeud, mesurée comme étant le nombre minimum de sauts depuis le noeud passerelle (gateway)
Le nombre d'interfaces radio disponibles sur un nœud
Notez que le nœud de la passerelle a le rang le plus élevé car il est prévu de réaliser le plus de trafic
Le rang pour les autres nœuds est donné par:
L'algorithme se résume comme suit :
I Ordonner tous les noeuds selon la formule
II Visitez chaque nœud dans l'ordre de classement (ordre décroissant)
II.1 Attribuer un canal à la liaison incident au noeud, si le nœud et l'un de ses voisins sont attribués un canal en commun
-Sinon: tandis que le noeud a un lien incident non assigné
II.2 Choisissez un voisin avec lequel le noeud a le plus de trafic dans la matrice de trafic
- Si le noeud a une radio non assigné
II.2.a Assigner à la radio le canal le moins utilisé dans le voisinage -sinon
II.2.b Allouez le lien un canal le moins utilisé dans ceux déjà attribuée aux radios du nœud
Since the algorithm is static and lacks node visit feedback, both execution time and complexity are minimized The sole metric employed is one that ranks nodes, facilitating the organization of node visits Consequently, the algorithm only addresses direct neighbor conflicts rather than the overall conflict problem.
2.3.2 CLICA: Connected Low Interference Channel Assignment
In [1], the authors introduced a polynomial-time frequency allocation heuristic called CLICA, which addresses two main constraints: network connectivity and minimizing conflicts [10] These constraints limit future assignment flexibility [1] Therefore, the core idea behind CLICA is to leverage this flexibility to allocate channels to nodes based on their priority levels Priorities are determined by the number of available radios at a node, necessitating CLICA to dynamically adjust the priorities assigned to channels [1][5].
Plus précisément, chaque noeud est associé à une priorité, et les décisions d'assignation sont réalisées noeud par noeud sur la base de l'ordre de leur priorité
Since priority changes are dynamic, initially, each node is assigned a default priority based on specific criteria such as its proximity to the gateway and traffic load.
As assignments progress, it becomes clear that the state of nodes and links changes, leading to reduced flexibility in allocating certain nodes, particularly those without free radio channels Consequently, the algorithm may overlook default priorities and instead focus on key constraints that elevate the priority of these nodes As a result, some nodes must utilize the channels allocated to their neighbors to maintain connectivity Conversely, if a node has flexibility in channel selection, allocation is performed using a greedy approach.
This algorithm aims to minimize interferences and conflicts by focusing solely on neighboring links However, it overlooks potential conflicts between non-neighboring links, which can arise based on the distance between their nodes As a result, this algorithm is not optimal.
The algorithm proposed in [7] is termed the "evolutionary greedy algorithm" due to its foundation in greedy algorithm characteristics It begins with an empty allocation, meaning that the links have no initial frequency The algorithm then identifies active disjoint links, where each node belongs to only one path, between the source and destination nodes Subsequently, the paths are ranked based on their potential capacity in descending order Frequency allocation occurs along the links of these paths, moving from the source to the destination, with the chosen frequency for each link being the one that maximizes the defined benefit function.
Paths : la liste des routes
B(i) est le débit du lien i sur un chemin j
IE(i) est le nombre de conflit du lien i (voir le calcul de MPI) [7]
This algorithm facilitates frequency assignment that maximizes capacity; however, it suffers from poor frequency reuse due to its metric, which considers links in conflict even when they are not (over three hops) We employ an optimized version of this algorithm that explores all possible assignments to select the best one Our approach includes accounting for traffic directionality in our initial experiments to assess the differences between directional and bidirectional conflicts.
Conclusion
In this article, we explored the various types of conflicts and examined algorithms designed to address frequency allocation issues The next chapter will focus on studying how traffic direction influences the occurrence of these conflicts.
Etudes et expérimentations
Evaluation théorique
To assess the differences in conflicts and gains between unidirectional and bidirectional traffic, we examine a topology consisting of eight nodes This analysis involves two scenarios: the first with unidirectional traffic and the second with bidirectional traffic Both cases will utilize the same number of available channels (f1, f2, f3) and radios We will determine the optimal allocation for each scenario and evaluate the number of conflicts present, as well as the overall capacity of each network, defined as the average number of kbits/s received by the destinations, assuming all links have the same capacity C.
Figure 3: Allocation optimal : cas directionnel
En considérant les propriétés proposés dans [7], ([d (E interférant, R) ≥ 3] et [d (E1, E2) ≥
To achieve optimal allocation, the distance between the interferer E and the receiver R must be at least three hops, as well as the distance between emitters E1 and E2 On the path (ACEGH), the links (A, C) and (C, E) are in conflict, as are the links (C, E) and (H, F), with the link (C, E) experiencing the highest number of conflicts, totaling three Similarly, for f3, the links (A, B) and (E, G) conflict, as do the links (E, G) and (G, H), with (E, G) also presenting three conflicts Therefore, on the path (ACEGH), the links (C, E) and (E, G) both have three conflicts, resulting in a maximum conflict count of three for this path Applying the same principle to the path (ABDFH) reveals a maximum of two conflicts According to formula (2.2), these findings highlight the critical role of conflict management in optimizing communication paths.
Figure 4: Allocation optimal : cas directionnel
Given that we have 8 active bidirectional links and need to allocate 3 frequencies, we must use 2 frequencies three times and 1 frequency twice to minimize conflicts Regardless of how frequencies are distributed across the links, those sharing the same frequency will inevitably conflict Consequently, the optimal allocation will always result in 3 conflicts on each path (ABDFH) and (ACEGH) Following the same principle as in the directional case, there will be a maximum of 3 conflicts on the path (ABDFH) caused by the link.
(A, B) and its competitors (A, C) and (F, H) are analyzed for frequency f1 Additionally, we identified a maximum of three conflicts on the path (ACEGH) generated by the link (G, H) along with its competitors (EG) and (BD) for frequency f3 Consequently, we have established the capacity.
Ces premières évaluations donnent un gain en capacité pour le cas directionnel par rapport au cas bidirectionnel car 0.83C > 0.66C mais ceux-ci sont des estimations théoriques.
Evaluation sur NS3
In this section, we will conduct evaluations using simulations on the NS3 simulator Similar to the theoretical case, we simulated frequency allocation for both unidirectional and bidirectional traffic, assessing the capacity gains in each scenario To achieve this, we modified the simulator to implement an evolutionary greedy algorithm that accounts for traffic considerations.
The execution time of algorithms significantly increased with the addition of nodes, leading us to limit our experiments to a maximum of 10 nodes This limitation hindered our ability to explore topologies with a larger number of nodes and obtain more conclusive results.
3.2.1 Paramètres de simulations et Résultats
Scénario 1 : Evaluation des la capacité par rapport au nombre de nœuds
In this scenario, we assess the reception and transmission capabilities in relation to the number of nodes across two traffic modes: unidirectional and bidirectional The key parameters impacting the experiments are summarized in the following table.
Type de trafic CBR (UDP) débit 11 Mbps
Figure 5: Evaluation de la capacité d’envoie par rapport au nombre de nœuds
The results indicate that with fewer than 6 nodes, both cases exhibit the same transmission capacity, as links can effectively share channels and avoid conflicts when more than 3 frequencies are available However, an improvement is observed in the bidirectional case between 6 and 8 nodes, attributed to the ability to utilize links in both directions, resulting in a higher volume of packet transmissions compared to the unidirectional case, which operates in only one direction Beyond 8 nodes, the advantages of the unidirectional case become apparent, as the performance curve rises while the bidirectional one declines This shift is due to the unidirectional case's ability to maintain fewer conflicts as new links are added To assess the impact of interferences and conflicts on each case, we evaluate the calculated reception capacity based on the number of packets received, as this metric accurately reflects the network's performance.
Figure 6: Evaluation de la capacité de réception par rapport au nombre de nœuds
When observing networks with fewer than six nodes, the reception capacity curves closely resemble the sending capacity curves, indicating that the number of packets received is nearly equal to the number of packets sent This balance is achievable because three frequencies are sufficient to eliminate conflicts in a topology with fewer than six nodes, while also preventing interference that leads to packet loss However, as the number of nodes exceeds six, the curves begin to decline due to increased conflicts resulting from the higher node count To determine which of the two traffic modes responds better to the increase in nodes, we calculated the ratio of received to sent packets to analyze which mode experiences more losses.
Figure 7: Evaluation de la capacité de réception par rapport au nombre de nœuds
This figure illustrates the relationship between received and sent packets, scientifically referred to as Packet Delivery Ratio (PDR) A PDR value close to 1 indicates good network performance As observed in previous figures, with fewer than six nodes, packet loss is minimal; however, losses begin to increase beyond six nodes Additionally, the directional case results in fewer losses compared to the bidirectional case, which shows a linear decline in its curve.
Scénario 2 : Evaluation des la capacité par rapport au nombre de canaux
In this scenario, we conducted the same evaluations while varying the number of available channels to determine if we would reach the same conclusions as when we varied the number of nodes.
Figure 8: Evaluation de la capacité d’envoie par rapport au nombre de canaux
The results indicate that with fewer than two channels, both traffic modes yield nearly identical outcomes However, a closer examination reveals that the directional case performs slightly better than the bidirectional case, suggesting higher transmission in the directional mode due to greater channel sharing in the bidirectional mode Starting from three channels, the bidirectional case demonstrates a more effective frequency distribution, resulting in increased traffic compared to the directional case, which operates in a single direction This leads to an enhancement in transmission capacity As in the previous scenario, we aim to assess the network's performance regarding these transmissions, particularly focusing on reception quality, which is influenced by interference conflicts.
Figure 9: Evaluation de la capacité de réception par rapport au nombre de canaux
The data indicates that directional receptions surpass bidirectional ones when there are fewer than two channels However, with more than three channels, bidirectional receptions increase This is attributed to the fact that bidirectional communication involves numerous transmissions due to links operating in both directions As additional channels are introduced, conflicts decrease, leading to an increase in receptions.
Figure 10: Rapport entre capacité de réception et d’envoie par rapport au nombre de canaux
To accurately assess network behavior, we calculated the ratio of receptions to transmissions Our findings indicate that with fewer than two channels, there are significant losses, particularly in bidirectional communication However, as the number of channels increases beyond two, the losses decrease, with unidirectional communication consistently experiencing fewer losses.
Conclusion
Evaluations indicate that in a directed topology, there are fewer conflicts and consequently lower losses compared to a bidirectional topology These experiments were based on established properties that utilize the number of hops as a metric to identify conflicts between links While this metric effectively minimizes conflicts, it may sometimes inaccurately classify nodes as conflicting when they are not In the following chapter, we propose an evaluation model that considers the distance between nodes as a metric for determining link conflicts, employing a stochastic model to represent the nodes' positions in space.
Modélisation Mathématique des conflits et Evaluation
Conflits directionnels
In this section, we estimate the probability of having k conflicts, meaning that k nodes are located within the same area defined by a circle of radius Dee, with directional links Once we determine this probability, we will use it to estimate the average capacity of the network.
Nous estimons la probabilité d’avoir des conflits dans une zone limitée par
La probabilité qu’il y ait k émetteurs en conflits :
La probabilité qu’un récepteur ait k émetteurs en conflits est:
Nous estimons ici le nombre moyen de conflits dans une zone limitée par avec les probabilités qu’un nœud soit émetteur ou récepteur
En considérant le cas de l’émetteur
au niveau de l’émetteur : E[Ke]= γ= (4.4)
au niveau du récepteur : E[Kr]= γ= (4.5)
We focus exclusively on one-hop communications, where the link bandwidth (BW) is assumed to be equal across all links This bandwidth is shared among all conflicting nodes, resulting in a maximum transmission rate at the sender of BW/K_e, where K_e represents the number of conflicting senders, and a maximum rate at the receiver of BW/K_r, where K_r denotes the number of conflicting senders with the receiver.
In a link formed between a transmitter and a receiver, the bottleneck is determined by the lowest throughput, whether at the transmitter or the receiver Therefore, we can assume that the overall throughput of the link will be dictated by this weakest point.
La moyen de C est E[C]= BW*E[min (
De plus P(min(X,Y)>x) = P(X>x ,Y>x) = P(X>x) P(Y>x) si X et Y sont indépendants
Avec la loi de poisson de paramètre E[Ke]= γ= , ó Dee est le rayon de détection de la porteuse
In a bidirectional scenario, a node functions as both a transmitter and a receiver This means that when a node transmits, it is also capable of receiving, and vice versa Therefore, the probability of transmission is greater than zero, while the probability of exclusive reception is zero.
We perform the same calculations as in the directional case, considering that the nodes have an equal probability of both emitting and receiving, denoted as Per The two nodes forming the link act as both transmitters and receivers, leading to a conflict zone that is defined as the intersection of their detection circles, represented by Dee This area is denoted as )=B(d,E1)U B(d,E1), as illustrated in the following figure.
Figure 12: Zone de conflits globale cas bidirectionnel
La probabilité pour qu’un nœud ait k conflit entant qu’émetteur :
La probabilité pour qu’un nœud ait k conflit entant qu’émetteur :
4.2.2 La moyenne des conflits bidirectionnels
En posant γ= et avec les mêmes calculs que pour les conflits directionnel nous aurons :
au niveau de l’émetteur : E[Ke]= γ= (4.10)
au niveau du récepteur : E[Kr]= γ= (4.11)
In a bidirectional traffic scenario, we analyze the conflicts between two emitters, defining the conflict zone as the union of the two circles created by the emitter nodes Consequently, we establish that C =
All conflicts arise from the nodes located within the area defined by two circles, each with a radius corresponding to the carrier frequencies of nodes E1 and E2 This area is represented as A(d, Dee), which is the union of the two circles formed by E1 and E2, denoted as B(d, E1) ∪ B(d, E2), where d is the distance between E1 and E2.
En partant du même principe que dans le cas directionnel, nous avons :
Avec la loi de poisson de paramètre E[ke]= γ= , avec Dee le rayon de la porteuse
, avec S l’aire de l’intersection de deux cercle et se calcule comme suit :
Pour deux cercles de rayon R et r, l’aire de l’intersection est
, donc l’aire de la zone de conflits est
Comparaison théorique
4.3.1 Rapport entre le nombre de conflits bidirectionnel et directionnel (en moyenne)
En moyen dans un lien nous avons : Pour les conflits directionnels E[Ke]= γ=
Et pour les conflits bidirectionnels E[Ke]= γ=
4.3.2 Distribution du nombre de conflits sur un lien bidirectionnel et directionnel
D’après les formules ci-dessus nous avons les courbes donnant la distribution du nombre de conflits directionnel et bidirectionnel:
Figure 13: Distribution des conflits directionnels et bidirectionnels : Equation (4.1)
Nous constatons qu’avec les mêmes paramètres, la probabilité d’avoir plus de conflits est plus grande dans le cas bidirectionnel que directionnel
4.3.3 Débit moyen dans les deux cas
En considérant le nombre de conflit moyen obtenu précédemment et la formule donnant le débit moyen sur un lien nous avons obtenu les courbes suivantes
Evaluation avec le nombre de k
Figure 14: Evaluation de la capacité dans les deux cas : Equation (4.7)
Nous avons un meilleur rendement en capacité dans le cas directionnel que le cas bidirectionnel
Figure 15: Evaluation de la capacité dans les deux cas : Equation (4.7)
The evaluation results indicate that as distances decrease, capacity increases This is because shorter distances result in fewer nodes within the conflict zone However, as the distance increases, the number of nodes rises, leading to more conflicts and consequently a decline in capacity.
Simulation sur NS3
In this section, we implemented a random node distribution model within a topology using the NS3 simulator We then evaluated the network's capacity behavior for both unidirectional and bidirectional traffic A set of nodes was randomly generated, with the number depending on a surface area framework of LxL, where L varies across different scenarios The distances between nodes were determined by a combination of a random number between 0 and 1 and a variable distance d, which was also adjusted in one of the scenarios In the unidirectional case, half of the nodes were designated as transmitters and the other half as receivers Similarly, in the bidirectional case, receivers were also capable of transmitting.
Type de trafic CBR (UDP) débit 2 Mbps
Tableau 2: Paramètres de simulations NS3: expérimentations finales
Scenario 1 : Evaluation des la capacité par rapport au délai
Figure 16: Evaluation des la capacité de transmission par rapport au délai
The results indicate that both unidirectional and bidirectional traffic exhibit nearly identical transmission capacity in relation to delay variations It is evident that transmission rates decrease as delay increases, which is expected since packets are transmitted less frequently under these conditions In the following figure, we will assess the reception capacity to determine which of the two traffic types responds more effectively to delay fluctuations.
Figure 17: Evaluation des la capacité de réception par rapport au délai
Directional traffic demonstrates superior reception capacity compared to bidirectional traffic, indicating significant losses in the latter due to congestion and interference from increasing conflicts It's important to note that the decline in performance is not solely due to losses; as delays increase, both transmissions and receptions decrease simultaneously The following figure will calculate the ratio of receptions to transmissions to assess the loss rate effectively.
Figure 18: Rapport entre la capacité de réception et la capacité de transmission par rapport au délai
The figure indicates that smaller delays result in significant packet losses However, as the delay increases, there is a slight decrease in transmissions, leading to an upward trend in the curves Notably, for delays greater than 0.01, directional traffic experiences fewer losses compared to bidirectional traffic This can be attributed to the fact that, despite the increased delay, bidirectional links are used in both directions, resulting in a higher number of packet transmissions lost due to congestion and interference To assess the impact of conflicts on network capacity, we will vary the distance between nodes, particularly the transmitters, to determine if increasing this distance reduces conflicts and to analyze the role of each traffic mode.
Scenario 2 : Evaluation des la capacité par rapport à la distance entre les nœuds
Figure 19: Capacité de transmission par rapport à la distance
We observe that by only adjusting the distance without changing any other parameters, bidirectional transmission outperforms directional transmission Additionally, bidirectional communication has the advantage of utilizing links in both directions, allowing both transmitters and receivers to send packets, unlike directional transmission, which does not operate in this manner.
We observed that at a maximum distance of 150 meters, there is no significant variation, indicating that the curves remain at the same level This suggests that the carrier distance exceeds 150 meters, causing the nodes to remain within conflict zones Given this observation, let's now examine the reception aspects.
Figure 20: Capacité de réception par rapport à la distance
Contrary to transmissions, increasing distance reduces losses, as we experience more receptions with greater distances Therefore, we can conclude that extending the distance helps minimize interference.
We consistently observe that directional receptions outnumber bidirectional ones, even though the latter offers significant transmission capabilities Consequently, it is directional traffic that most benefits from increased distance while minimizing interference.
The two previous figures provide a clear understanding of the relationship between receptions and transmissions; however, we present this relationship below to highlight the actual differences.
Figure 21: Rapport entre la capacité de réception et la capacité de transmission par rapport à la distance
As expected, increasing distance results in fewer losses for directional signals compared to bidirectional ones However, we observe that starting from 80 meters, there is an improvement in bidirectional performance At this distance, reduced interference enhances the reception capacity.
Our study primarily aims to examine the impact of traffic modes on conflicts and the sharing of the medium To achieve this, we will vary the number of nodes, as they are the entities that share the channel The following scenario highlights this experiment.
Scenario 3 : Evaluation des la capacité par rapport au nombre de nœuds
Figure 22: Capacité de transmission par rapport au nombre de nœuds
The transmission capacity increases with the number of nodes, which is a logical observation Importantly, both curves exhibit similar trends, indicating that a higher number of nodes leads to an increase in the volume of packets transmitted Notably, the bidirectional case consistently shows more transmissions than the directional case To further understand the impact of increasing the number of nodes, we can examine the reception rates illustrated in the figure below.
Figure 23: Capacité de réception par rapport au nombre de nœuds
Nous remarquons toujours que il y a plus de réceptions en directionnel qu’en bidirectionnel
The increase in nodes has a lesser impact on conflict presence in directional cases compared to bidirectional ones Initially, the curves decline but begin to rise after reaching 25 nodes It's important to note that this does not indicate a reduction in conflicts; rather, as the number of nodes increases, conflicts also rise The growth in transmission leads to an increase in receptions, yet losses continue to escalate with the number of transmissions This will be further illustrated below.
Figure 24: Rapport entre la capacité de réception et de la capacité de transmission par rapport au nombre de nœuds
As the number of nodes increases, the ratio of receptions to transmissions significantly decreases due to a high incidence of losses caused by congestion and interference, leading to conflicts Initially, with fewer than 16 nodes, the directional case experiences fewer losses compared to the bidirectional case However, as the number of nodes rises, the topology begins to resemble that of the bidirectional case in terms of conflicts, resulting in both cases experiencing the same loss rate.
In this section, we examined the impact of traffic patterns by simulating a random distribution model of nodes in space Our observations indicate that bidirectional traffic enhances transmission rates; however, it also leads to higher loss rates compared to unidirectional traffic These losses can be attributed to congestion as well as conflicts within the network.
Conclusion Général et perspectives
Conclusion général
In this study, we examined the channel assignment problem in mesh networks, focusing on the impact of traffic direction on conflict presence We utilized the algorithm proposed in [7] to incorporate traffic direction for our initial experiments Subsequently, we developed a mathematical model that accounts for the random distribution of nodes within a topology to assess this impact Finally, we implemented this random node distribution in the NS3 simulator to conduct our final experiments.
Our experiments indicate that traffic direction significantly impacts conflict presence and overall network capacity We found that directional traffic facilitates better channel reuse compared to bidirectional traffic, leading to increased long-term reception throughput within the network However, bidirectional traffic allows for a higher volume of transmission by utilizing links in both directions.
The distinction between unidirectional and bidirectional traffic in the presence of conflicts depends on specific cases related to the number of links and frequencies In certain scenarios, both traffic types may experience the same number of conflicts based on the quantity of nodes and available frequencies However, in such cases, bidirectional traffic can facilitate higher transmission rates by utilizing links in both directions When the network has sufficient bandwidth to reduce congestion, bidirectional traffic can significantly enhance the overall capacity of the network.
Perspectives
Optimizing the complexity of optimal allocation algorithms is crucial for reducing allocation time Proposing an algorithm that considers traffic patterns to leverage the advantages of each is a viable approach.
The algorithm used in our initial experiments proved to be highly optimal; however, execution time significantly increased with the number of nodes This limitation led us to conduct experiments with fewer nodes, highlighting the need for a high-capacity computing platform to perform simulations across multiple nodes Additionally, it is crucial to carry out real-world experiments to assess the impact of traffic direction in a practical environment.