Skip to content Skip to footer

Sealing the Afghan-Pakistan Border – A Case Study

Author Profile: Syed Aseem Ul Islam is a Research Scholar at the University of Michigan, Ann Arbor, USA, specializing in adaptive and model-predictive flight control systems. He received his bachelor’s degree in aerospace engineering from the Institute of Space Technology, Islamabad, and his master’s and Ph.D. degrees in flight dynamics and control from the University of Michigan.

With the withdrawal of the United States from Afghanistan imminent there is a serious concern among Pakistani planners about the effects of the situation in that country spilling over into Pakistan. 

The concerns are two-fold: First, preventing terrorist groups that may find safe haven in Afghanistan from crossing the border into Pakistan. Second, preventing an influx of refugees in response to a possible civil war in Afghanistan. 

This is made clear by Prime Minister Imran Khan’s interview to the New York Times where he is quoted to have said:

“What if [the] Taliban try to take over Afghanistan through [the] military? Then we will seal the border, because now we can, because we have fenced our border, which was previously [open], because Pakistan does not want to get into, number one, conflict. Secondly, we do not want another influx of refugees.”

It is becoming increasingly likely that we will see a Taliban takeover of Afghanistan, followed by civil war. 

Of course, the Pakistani state has been aware of this possibility and has made efforts to mitigate its effects. The much-touted border fence is mostly complete, with statements that it would be completed by June 30th

Ignoring the improbability of the fence being completed so quickly, a border fence on its own cannot “seal” the border. The border fence (like any other barrier) can only slow someone from crossing. For this reason, forts have been planned along the fence. 

To investigate the dynamics of an illegal border crossing, let us assume some parameters. 

Firstly, the fence must be cut or overcome in some way. Let us consider that this process can be done as quickly as 10 minutes by a motivated crosser. 

Secondly, the border will be crossed on foot or with assistance by animals or off-road vehicles, and thus, the crosser will enter Pakistan at an average speed of 40 km/h. 

This scenario gives us a time of 15 minutes before the person crosses 10 km inside Pakistan (and thus, will be too difficult to find, track, and intercept) 

To stop an illegal border crossing, we must detect the crossing within 10 minutes and intercept the individuals that have crossed 15 minutes after they have crossed the border.

Using ground-based forts, the defenders are subject to the same mobility constraints as the crossers – they must travel on poorly built roads or off-road at average speeds of 40 km/h. 

With the planned construction of 443 forts, interception of crossers should not be an issue as each fort will manage a border length of roughly 6 km (assuming the entire length of the Durand line even though the fence does not span the entire Durand line). 

However, detection of crossings will be a serious capability gap as ground-based surveillance systems cannot see distances as long as 6 km over uneven. The solution to this gap is obvious: unmanned aerial vehicles (UAVs) and this article will provide a case study for surveilling the AfPak border using UAVs.

The Border

Figure 1: AfPak Border divided into blue and red zones. Blue zone is served by Mianwali AFB, red zone is served by Samungli AFB.


We will consider the worst-case scenario and include the entire length of the Durand line for surveillance. We will divide the border into two regions: blue and red. 

The blue region will be served by UAVs flying out of Mianwali AFB, which is 147 km from the border. 

The red region will be served by UAVs flying out of Samungli AFB, which is 57 km from the border. 

We will assume that the UAV bases are on average 100 km from the border. 

Since we are dividing the border into two equal parts, we will only consider one half, and the lessons from it would apply equally well to the other half. Considering the Durand line is 2,670 km long, each half is 1,335 km long.

The mission profile for the UAVs will consist of:

  1. Taking off from the AFB, climbing to cruise altitude, while flying straight towards the border, then either turning 90 left or right.
  2. Flying along the border till the end of the blue or red region.
  3. Changing the bearing by 180 and flying along the border again and returning to the location where the border surveillance was started.
  4. Return to base for landing.

The flightpaths for the red region are visualized below:

Figure 2: Flightpaths out of Samungli AFB for the red region.


Note that the lengths of the green and orange flight paths are smaller than the maximum flight path possible with the UAV that we select below.

The UAV: Shahpar-II

We will propose a capability built upon the recently inducted Shahpar-II UAV for the reason that being an indigenous system, it will be cheaper and faster to produce and induct in numbers. 

For specifications on the Shahpar-II we refer to the display at IDEX 2021:

Figure 3: Shahpar-II at IDEX 2021.


We will assume that Shahpar-II can cruise at 17,000 ft (3,000 ft less than the ceiling), which translates to approximately 5,200 m. Furthermore, we will assume a cruising speed of 130 kts (reasonable for a UAV of this size), which translates to approximately 66 m/s. 

Using the maximum endurance of 14 hours and assuming an average speed of 45 m/s, we can conclude that a Shahpar-II can fly a mission distance of 2,268 km. This is large enough that this limit will not play a role in the case study that we will pursue. 

One possible concerning aspect is the stated data-link range of 300 km. However, the presence of a hump point towards the presence of a SATCOM antenna, thereby enabling an unlimited data-link range. 

The SATCOM feature may not have been advertised at IDEX 2021 to avoid issues with MTCR’s 300 km limit on UCAVs. Therefore, we will assume that a SATCOM data-link is available for use by Pakistani armed forces.

Next, we will assume that the locally manufactured Zumr-1(EP) is the primary imaging payload on the Shahpar-II. Fortunately, specifications for this system can be found online. Of interest to us are the specifications of the thermal imager:

Figure 4: Specifications of the thermal imager on the Zumr-1(EP) imaging payload.


We will assume that the wide field-of-view (FOV) can be used to detect possible points of interest, while the narrow FOV can be used to confirm any detections. 

The cruise altitude, together with the wide and narrow FOV translates to 0.3 and 6 pixels per meter, respectively. We will also assume that the crosser is stationary, as they will be while attempting to break or otherwise overcome the border fence. 

Putting all this information together leads us to the following geometry:

Figure 5: Ground visibility for a UAV cruising at 5,200 m (not to scale).


A Shahpar-II cruising at 5,200 m, using its Zumr-1(EP) payload’s wide FOV thermal imager, can look at a region 2,200 m across. 

Next, a stationary target will remain in the wide FOV for a maximum of 33 seconds based on the cruise speed. Finally, the narrow FOV needs to rotate at a rate of  0.7/s only, which is entirely realistic. 

Therefore, we can conclude that a single Shahpar-II can “take care” of a border length of 2,200 m. However, this is not the entire story.

Next, we will consider multiple Shahpar-II’s flying uniformly apart, up and down along the border. The geometry of this setup can be drawn in two configurations shown below:

Figure 6: Geometry for multiple Shahpar-II’s flying up and down along a straight line, with target shown as an orange cross (not to scale).
Figure 7: Geometry for multiple Shahpar-II’s flying up and down along a straight line, with targets shown as orange crosses (not to scale).


Assume that the distance between UAVs x is much greater than 2,200 m. Consider Figure 6, which shows that the target will be detected in 2,200/66=33 s by the UAV flying towards the left. 

Next, consider Figure 7, which shows that for a target in the center of the red blind region it will be detected by both directions of drones in x-44002×128 s, as the red blind window is shrinking at the sum of the speeds of the drones, which is 66+66=128 m/s. 

Note that Figure 7 provides the worst-case scenario, and thus gives us an estimate of the maximum time a crosser can remain undetected for UAVs flying x m behind each other in a loop along the border.

Using the length of 2,670 km length for the Durand line, we can compute the number of UAVs required in the air at the same time to ensure a certain “maximum blind time,” which is the maximum amount of time a stationary target can remain undetected. 

Figure 8 plots the maximum blind time in minutes versus the number of UAVs simultaneously in the air on the entire border. We can see that a maximum blind time of under 20 minutes can be achieved with as few as 18 UAVs, whereas a maximum blind time of under 10 minutes requires 36 UAVs.

Figure 8: Maximum Blind Time versus Number of UAVs for the entire Durand line. Using 36 UAVs a maximum blind time of under 10 minutes can be achieved.


The Cost

In order for planners to make a decision on such a border surveillance plan, the most important factor is cost.

Since procurement and operational costs of Shahpar-II are not available, we will make estimates based on similarly sized UAVs. 

We will assume a unit cost of $0.7 million and an operating cost of $700 per hour. Furthermore, we will assume an availability rate of 33%. That is, to fly 10 aircraft simultaneously, we will require 30.

Figure 9: Procurement and operating costs for the entire border surveillance program versus maximum blind time in minutes.


Using Figures 8 and 9 we can present three options for planners:


Noteworthy is the fact that the majority of the costs are incurred by operating costs, as a large fleet of UAVs would be required to fly 24 hours a day, 365 days a year. 

MATLAB code used in the presented case study can be made available on request. This code can be used to explore additional options. For example, UAVs other than the Shahpar-II may be considered that will cruise at different altitudes and speeds. 

Furthermore, the length of the border that requires aerial surveillance may be modified since it is unrealistic to surveil the entire Durand line, given the fact the many near-flat regions can be effectively monitored using ground-based devices.


It can be seen from the table above that operating a fleet of Shahpar-IIs for nonstop border surveillance gets increasingly expensive as the maximum blind time is reduced. 

This case study is not meant to suggest that such an option is perfect for Pakistan. The intention of this case study is to provide ballpark numbers for near-nonstop border surveillance using UAVs. 

This may dissuade some readers who thought that this was a viable option before. On the other hand, it may convince skeptical readers of this option’s viability.

Show CommentsClose Comments

Leave a comment